Intelligent Robotics and Applications

In recent years, the new generation of information technology has been widely applied in manufacturing domain. Building intelligent workshop and achieving intelligent manufacturing have become the purposes of industry development. The current workshop service systems just fulfill the mapping between physical and digital layer, while have not completed the interconnection and interaction between physical world and information world. It is the emergence of digital twin that become one of the solution to this bottleneck. In this paper, a system framework of digital twin workshop is proposed. In the light of the framework, a model of digital twin workshop based on physical perception data is established. This model is divided into three parts, which including workshop physical model, digital model based on ontology, and virtual model. Moreover, the operation mechanism of digital twin workshop is described among the three models. Finally, a three-dimensional model for a production line is built, and the connection between digital and virtual layer is established and demonstrated.

This paper presents a new stable factor approach of input-output-based discrete-time sliding-mode control (IODSMC-SF), which dedicates to piezoelectric actuators (PEAs) with non-minimum phase (NMP) property. This control approach is developed based on a linear discrete-time input-output nominal model. A stable factor, which ensures stable and accurate motion control for PEAs with NMP nature, is designed, analyzed and introduced into the controller. One unique feature of the proposed controller lies in that it ensures stable and precision motion control for PEAs with NMP property. The construction of either a hysteresis model or a state observer is not needed. Moreover, the proposed controller releases the burden on parameter selection since only a stable factor is needed to stabilize the NMP system and this factor can be obtained by optimization algorithm. Experimental results with a piezoelectric actuator are presented to demonstrate the effectiveness of the proposed controller.

This paper designs the dynamic system and main control circuit of quadrotor unmanned aerial vehicle (UAV). Power system of the quadrotor UAV includes brushless DC motor (BDCM), matching paddle, power lithium battery and drive circuit of (BDCM). Its main control hardware circuit includes the main controller minimum system, the sensor module, the wireless communication module and the power module. The attitude algorithm and attitude control algorithm are designed. The Kalman filter algorithm is used to calculate the attitude of quadrotor UAV. The cascade PID control algorithm is used to control attitude. Finally, a quadrotor UAV prototype was trial-produced, and the tested results show that the aerial vehicle meets the design requirements.

A real timework-shop dust detection system based on Single-Chip Microcomputer was designed. Aiming at complexity of manual configuration for MS3110, the concept of self-correcting was proposed and realized. The system provides RS232 communication and Bluetooth wireless communication mode. Comparing with the traditional detecting devices, it can greatly shorten the time consuming of detection as ensuring the measuring accuracy.

Rehabilitation robot has direct physical interaction with human body, in which the adaptability to interaction, safety and robustness is of great significance. In this paper, a compact rotary series elastic actuator (SEA) is proposed to develop an elbow rehabilitation robot for assisting stroke victims with upper limb impairments perform activities of daily living (ADLs). The compliant SEA ensures inherent safety and improves torque control at the elbow joint of this rehabilitation robot. After modeling of the rotary stiffness and dynamics of the SEA, a PD feedback plus feedforward control architecture is introduced. A test bench has been designed to experimentally characterize the performance of the proposed compliant actuator with controller. It shows an excellent torque tracking performance at low motion frequency, which can satisfy the elbow rehabilitation training requirement. These preliminary results can be readily extended to a full upper limb exoskeleton-type rehabilitation robot actuated by SEA without much difficulty.

A vibro-tactile stimulation and vibro-signature synchronization device was proposed in this paper. The device was used to elicit steady-state somatosensory evoked potential (SSSEP) and synchronize the actual vibraion information with the corresponding electroencephalography (EEG) signals in temporal domain. The device provided five independent stimulation channels and could generate vibro-tactile stimulation with arbitrary waveform, amplitude and frequency. Each channel used the intended stimulation waveform to drive a linear resonant actuator (LRA). The vibro-signature of each channel could be detected by a force sensing resistor (FSR) and fed back into the EEG recording system. Four stimulation patterns with a random sequence were applied to the index finger of three healthy subjects. Results showed that SSSEP features could be evoked with different stimulation patterns and the actual vibro-signature could be synchronized with the EEG signals. Comparing the actual stimulation with the intended stimulation, the errors of carrier frequency and modulation frequency were quite small. Results indicated that this novel device stood a good chance of serving in SSSEP-based studies.

When it comes to indoor positioning system (IPS) using smartphones, most of pedestrian dead reckoning (PDR) based IPS systems rely on the embedded sensors in a smartphone. Unfortunately, those sensors cannot avoid having noisy data due to their poor performance. Unpredictable indoor environments and a user’s motion may also distort sensor data. For these reasons, an error-compensating algorithm is required. In this paper, we propose a compensated gyroscope sensor algorithm to clear away those noises. Also, we used Bluetooth-Low-Energy (BLE) beacons based positioning system. It aims for reducing the compensated gyroscope sensor’s cumulative error as well as providing precise user’s initial point. Finally, we propose an integrated system making cooperative relationship between BLE beacons and sensors, especially compensated gyroscope sensor. This paper also includes implementations in the hallway of a building to describe the effectiveness by showing error rates.

Current electrophysiological monitoring is based on invasive electrodes or surface electrodes. Here, a surface electromyography (sEMG) electrode with self-similar serpentine configuration is designed to monitor biological signal. Such electrode can bear rather large deformation (such as >30%) under an appropriate areal coverage. And the electrode conformally attached on the skin surface via van del Waal interaction could furthest reduce the motion artifacts from the motion of skin. The capacitive electrodes that isolates the electrodes from the body also provide an effective way to minimize the leakage current. The sEMG electrodes have been used to record physiological signals from different parts of the body with sharp curvature, such as index finger, back neck and face, and they exhibit great potential in application of human-machine interface in the fields of robots and healthcare. Integrating wireless data transmission capabilities into the wearable sEMG electrodes would be studied in future for intelligent could healthcare platform.

A stiffness decoupling 8/4-4 parallel force sensing mechanism (PFSM) is presented. Its mathematic model is established with screw theory. The force mapping relation is studied and the stiffness matrix is found to be a diagonal matrix, which proves the stiffness decoupling characteristics of the mechanism. According to the concept of fully isotropy, the isotropy conditions are analyzed, the parameters which meet fully isotropy are given. The 8/4-4 PFSM’s configuration under isotropy parameters is analyzed. Based on this configuration, an 8/4-4 mechanism cluster which meets the fully isotropy is presented. The cluster’s configuration is classified and induced into four main configurations according to the different parameter conditions.

The application of pattern recognition techniques to Electromyography (EMG) signals has shown great potential for robust, natural, prostheses control. Despite promising development in EMG pattern recognition techniques, the non-stationary properties of these signals may render these techniques ineffective after a period of time, subsequently demanding frequent recalibration during long term use. Potentially one method to reduce the impact of non-stationary traits of EMG signals is through attempting to construct a training dataset that represents this gradual change in the signal. In this paper, we investigate the potential impact of data selection schemes for inter-day motion recognition, across a period of five days of high density data recording with an LDA classifier, and present our preliminary findings. This paper proves that training a classifier with data from several spaced points of a single day can improve its inter-day performance which subsequently supports the long term use of prosthesis. Therefore the work presented here may aid in furthering our understanding of the physiological changes in EMG signals and how they may be exploited to further improve the robustness of pattern recognition methods for long term use.

Aiming at the limitation of the traditional flexible robot’s single adjustment stiffness or damping, a magnetorheological (MR) actuator of which stiffness and damping can be adjusted simultaneously and independently is proposed for the robot joint. The principle of equivalent variable stiffness and damping is analyzed theoretically, and the adjustment range of stiffness and damping is deduced. As the first step, the performance of variable damping is evaluated with experiment by using a MR damper. The preliminary results show that the magnetorheological actuator is capable of changing the damping by controlling the current applied to the damper.

This paper presents a method utilizing temperature field to improve the indoor mobility of the blind/visually-impaired-people (Blind/VIP) in a physical field-enhanced intelligent space (iSpace) that takes advantages of the rapidly developing cloud-computing and personal mobile devices (generally built with sound, image, video and vibration alert capabilities) to share way-finding information among users. A method, which uses temperature fields and their gradients to detect face-orientation and analyze leg-postures for predicting the motion states of other humans in a traveling path, is introduced. The concept feasibility of the temperature-based human motion detection has been experimentally validated in a simulated school environment for the Blind/VIP where users are generally familiar with stationary objects but less confident in daily walking in a crowd where human motion is unpredictable. The experimental findings presented in this paper establish a basis for developing temperature filed-enhanced iSpace.

Motion capture of human body potentially holds great significance for exoskeleton robots, human-computer interaction, sports and rehabilitation research. Dielectric Elastomer Sensors (DESs) are excellent candidates for wearable human motion capture system because of their intrinsic characteristics of softness, lightweight and compliance. Fabrication process of the DES was developed, but a very few of optimal design is mentioned. To get greater measurement precision, in this paper, some optimization criteria was put forward and validated by some experiments. As a practical example, the sensor was mounted on the wrist to measure joint rotation. The experiment results indicated that there is a roughly linear relationship between the output voltage and the joint angle. Therefore, the DES can be applied to motion capture of human body.

Aiming at the problem of person tracking for mobile robot in complex and dynamic environment, a multi-feature tracking strategy is proposed in this paper, by which the target can be determined based on the joint similarity. The joint similarity consists of motion model similarity, color histogram similarity and human HOG feature similarity. The tracking of target is realized by the method of joint likelihood data association. The above strategy can solve the problems such as similar color interference, target loss, and target occlusion. In addition, considering the lost target, a fast search strategy is proposed to search the target. Finally, the method is tested with the mobile robot. The experimental results show that the proposed method is robust and effective when the target is moving rapidly, and it can satisfy the real-time requirement of the system.

In this paper, a method of biped walking on a slope is studied, taking the NAO humanoid robot as research object. Firstly, we adopt the 3D linear inverted pendulum mode (3D-LIPM) to generate walking pattern, and obtain the reference trajectory of the center of mass (CoM). Then the Denavit-Hatenberg (D-H) parameters of the leg chain are determined based on the analysis of the NAO leg configuration. Finally in the process of walking, an extended Kalman filter (EKF) via fusing sensor data is used to estimate the robot’s CoM motion, in addition, an inverse kinematics (IK) controller is implemented which regulates the CoM position state in real time based on position tracking errors. On a slope with given angle, walking uphill and downhill experiments are conducted. The experimental results show that for the NAO humanoid robot, the deviation of walking direction can be controlled within 2 cm, so that it can keep walking stability for a long distance.

The Simultaneous Localization and Mapping problem limits the promotion of home cleaning robots in practical domestic environments. In this paper, a novel topological-vector based simultaneous localization and mapping (TVSLAM) algorithm is proposed to solve the problem. The algorithm involves four aspects. First, the ultra-wideband localization and dead reckoning localization are selected to develop a new combined localization algorithm which can improve the localization accuracy. In addition, a data acquisition algorithm which simplifies the process of data collection and demands much smaller memory size is proposed. Furthermore, a partitioning algorithm is developed to adapt to the various change rates of different rooms. Finally, an autonomous learning algorithm based on the regular and repetitive cleaning task is put forward. It makes the constructed map approach to the real environment with the increase of cleaning times. Overall, a novel topological-vector map is generated according to the above process of the algorithm. Simulation results show that the TVSLAM is an efficient and robust localization and mapping algorithm.

This paper presents a wall-climbing robot called Vortexbot, which has a suction unit that uses vortex flow to generate a suction force. Unlike the traditional unit based on contact-type adsorption, the suction unit does not touch the wall surface, which greatly reduces the frictional resistance between the robot and wall and improves the passing ability of the robot. It first introduces the principle of the vortex suction unit. Then, the authors design the mechanical structure of Vortexbot. Furthermore, they survey the suction properties of the suction unit on a smooth wall surface. In addition, they study the effect of the roughness and shape (a raised obstacle and groove) of the wall surface on the suction performance of the suction unit. Finally, they experimentally verify the climbing performance of Vortexbot on several kinds of walls with different surface conditions.

To allocate welding tasks to multiple robots and find collision free paths for them, an approach of multi-robot motion planning and simulation based on genetic algorithm is proposed. Priorities of welding points are defined based on the sequence constraints of welding points. Welding points are allocated to multiply robots and the objective is to minimize the welding time of station. Adapted genetic algorithm is proposed to seek the optimized solution. The three dimension model of welding assembly line is built in eM-power software. The welding robots move along the allocated welding points in the virtual environment, which can find and settle collisions between two robots or between robot and parts, and free collision path is founded finally. Welding path simulation can sharply shorten the planning time for the task allocation of multiply robots.

This paper presents a leader-follower formation integrated control method based on artificial potential field (APF) and sliding mode control (SMC) in an unknown environment with obstacles. Firstly, the online path planning in formation control is executed via APF to find a collision-free path for leader from the initial position to the goal position. Then, the trajectory tracking controller is designed via SMC method to adjust the linear velocity and angular velocity of the followers to form and maintain the predefined formation. Finally, the effectiveness of the proposed formation integrated control method has been verified by simulation.

In this paper, the issue of trajectory tracking for a non-holonomic three-wheeled mobile robot is researched and a controller with two layers’ structure which separately deals with the kinematic and dynamic characteristics of the mobile robot system is constructed. For the kinematics, a conventional and common kinematic controller is chosen to transform the position tracking errors into command velocity which will be taken as the reference input into the proposed dynamic controller. For the dynamics, a terminal sliding mode control strategy is designed to track the command velocity generated by the kinematic controller. The stability of the proposed control scheme has been proved by Lyapunov theory. Practical experiments are carried out to verify the effectiveness and accuracy of the proposed control algorithm.

Apply ball-picking robot in the golf, table tennis, tennis and other small-ball training court can enhance the efficiency of picking up, reduce the operating costs of the stadium, and improve the level of technical service. This paper explores the research of ball-picking robot. Analyzes the research results and application of key technologies in this field, such as navigation and location, path planning and wireless communication. Discusses the trend of intelligent ball-picking robot technology under the background of rapid development of intelligent core technology such as Internet, multi-robot cooperation, machine visual and multi-sensor fusion. The results show that, in addition to golf-picking equipment, the current ball-picking robot technology is still in the primary test stage. The degree of intelligence is still relatively low. There are few achievements can be popularized and applied. The industrial robot in recognition and localization, autonomous navigation, artificial intelligence have a rapid development. They will promote the research on ball-picking intelligent service robot technology. And promote the development of entertainment fitness industry.

Many researches have studied indoor localization techniques in past decades. Depending on a wireless sensor network, current distance-based localization techniques exploit different measurements of Received Signal Strength (RSS) values between RF devices. It is simple to implement and costly efficient, while the estimation accuracy is significantly reduced in indoor environments. In this paper, we focus on localization methods in real indoor environments using IEEE 802.15.4 standard Zigbee network. The objective is to mitigate the instability and divergence of signal strength by using successional RSS evaluations with Kalman filtering and avoid multiple NLOS interferences by using LOS node set identification procedure.

The indoor localization relies on the IMU sensor on the smartphone because there is no signal like GPS that compensates for location in real time like the outdoor. However, if IMU sensors are not periodically reset, the localization error will increase. Although various methods are suggested, it is difficult to apply it to any indoor location because there are few methods considering the spatial characteristics of the indoor. In this paper, we divided the indoor characteristics into two parts and suggested an integrated system considering the spatial characteristics.

This paper presents a characterization of the Sick LMS511-20100Pro laser range finder. With high accuracy and good robustness, this range sensor is suitable for various mobile robotic applications both indoors and outdoors. However, very few studies concerning the performance characterization of the LMS511-20100Pro can be consulted for better understanding and utilizing this sensor in practice. Therefore, some factors that could influence the sensor performance, such as drift effect, target distance, angular resolution, target material and mixed pixel, were tested and further analyzed. The effect of target distance and material on intensity information was also investigated. All the performed experiments demonstrate that the performance of the LMS511-20100Pro exceeds the specified values and can meet the requirements of simultaneous localization and mapping, although its practical performance might be significantly affected by some certain factors.

Nowadays there are a lot of kinds of cleaning robots which producted by different manufacturers come into people’s lives. But it is still a problem that how to evaluate each robot’s performance to check whether the quality is acceptable. In this paper, we make the first trial to evaluate the complete coverage path planning algorithm which is the core algorithm of a cleaning robot with Mocap system, and three simple metrics were proposed to evaluate overall performance of the algorithm. Lastly, the comparisons between different kinds of robots are presented.

Calibration is important to service robot, but the process of calibration is time consuming and laborious. With the popularity of service robot, an automatic and universal calibration system is urgent to be developed, therefore we propose a general batch-calibration framework, Motion Capture System is adopt as an external measurement device in virtual of it can provide realtime, accurate movement data of measured objects. We will show that the system is effective and promising with a case study of odometry calibration.

In this paper, we describes a novel proposal for the autonomous navigation control of quadrotor micro aerial vehicles for trajectories tracking in the XY plane. The quadrotor vehicle is an AR.Drone 1.0 from the company Parrot with a nonlinear behavior. The proposal includes system modeling, controller design, planning and simulation of the results. In our approach, we separate the model into two primary models: A linearity for the steady state and a nonlinearity for the dynamic transition.

We present a approach to real-time localization and mapping using a RGB-D camera, such as Microsoft Kinect, and a small and powerful computer Intel Stick Core M3 Processor. Our system can run the computation and sensing required for SLAM on-board the UGV, removing the dependence on unreliable wireless communication. We make use of visual odometry, loop closure and graph optimization to achieve this purpose. Our approach is able to perform accurate and efficient on-board SLAM, and we evaluate its performance thoroughly with varying environments and illumination conditions. The experiments demonstrate that our system can robustly deal with difficult data in indoor and outdoor scenarios.

Path planning is a beneficial strategy in improving the robustness of uncalibrated visual servoing. In this paper, a novel controller based on projective homography along with a new expression of optimal camera path are proposed to accomplish visual servoing tasks under totally uncalibrated situations. The interpolated poses of optimal path is express in the current planned frame, and the expression of optimal path is completely free of camera parameters and homography decomposition. The tracking of planned path is realized in projective homography space directly, based on the novel uncalibrated controller denoted as PHUVS. The simulation results prove the effectiveness of the proposed path optimization method and PHUVS controller.

This paper proposed a Fully Cloud-based Modular (FCM) Robot technology that enables almost all of the business logic program of the home service robot running on the cloud and the robot execute the order locally on control aspects. Moreover, its modular feature can reduce the cost of the robot and, meanwhile, provide a flexibility of customization, upgrades, maintaining. The traditional home service robot often uses cloud computing algorithms and cloudDB, but the basic calculations have been solved locally, which leads to a condition that the robot have to carry many additional facilities. Some concepts of robotic intelligence system have been put forward in the paper including: integrating single-board computers and microcontrollers as the distributed computing unit on the robot, a cloud server structure for FCM robot, finite-state machines and network protocol for solving common problems caused by fully depending on the cloud service. Finally, we built a home service robot following the concept of FCM technology. Comparing with traditional service robots, several experiments are conducted to verify the performance of the FCM technology.

In this paper, we introduce a novel scheme for tracking moving person based on particle filter and social force model. The tracking process contains two parts: the predict model and the decision model. We adopt the particle filter algorithm to predict the position and velocity of human. According to the result of prediction, we adapt a sophisticated motion model to calculate the value of social force. Finally, we can control the velocity of robot dynamically through the value of social force.

In order to improve the efficiency and trajectory tracking accuracy of a robot and reduce its vibration, this paper uses the sequential quadratic programming (SQP) method to perform time-jerk (defined as the derivative of the acceleration) optimal trajectory planning on a 7-degrees-of-freedom (DOF) redundant robot. Kinematic constraints such as joint velocities, accelerations, jerks, and traveling time are considered. When utilizing the SQP method, the initial input is set as average time intervals, and the output is optimal time intervals. Trajectory planning simulations in joint space are performed with optimal time intervals, the results showed that the SQP method is effective and feasible for improving working efficiency and decreasing vibration.

This work presents kinematic modeling and a kinematic nonlinear controller of an omnidirectional mobile platform that generates saturated reference velocity commands for path following problem. The dynamic compensation controller is considered through of a platform-inner-loop system to independently track four velocity commands. Stability and robustness of the complete control system are proved through the Lyapunov method. Finally, simulation results are presented and discussed, which validate the proposed controller.

In this paper, we proposed a method for plant classification, which aims to recognize the type of leaves from a set of image instances captured from same viewpoints. Firstly, for feature extraction, this paper adopted the 2-level wavelet transform and obtained in total 7 features. Secondly, the leaves were automatically recognized and classified by Back-Propagation neural network (BPNN). Meanwhile, we employed K-fold cross-validation to test the correctness of the algorithm. The accuracy of our method achieves 90.0%. Further, by comparing with other methods, our method arrives at the highest accuracy.

Viewpoint independent 3D object pose estimation is one of the most fundamental step of position based vision servo, autopilot, medical scans process, reverse engineering and many other fields. In this paper, we presents a new method to estimate 3D pose using the convolutional neural network (CNN), which can apply to the 3D point cloud arrays. An interest point detector was proposed and interest points were computed in both source and target point clouds by region growing cluster method during offline training of CNN. Rather than matching the correspondences by rejecting and filtering iteratively, a CNN classification model is designed to match a certain subset of correspondences. And a 3D shape representation of interest points was projected onto an input feature map which is amenable to CNN. After aligning point clouds according to the prediction made by CNN, iterative closest point (ICP) algorithm is used for fine alignment. Finally, experiments were conducted to show the proposed method was effective and robust to noise and point cloud partial missing.

CNC milling is widely used in manufacturing complex parts of aerospace fields, and the development of the intelligent tool wear monitoring can improve the utilization of the tool during the milling process while ensuring the surface quality of the processed parts. In this paper, a novel method based on wavelet packet analysis and RBF neural network was proposed for monitoring the tool wear condition during milling. Firstly, cutting force signals were measured during milling, and filtered by filter function. Secondly, the cutting vibration signals caused by tool wear were separated by the wavelet packet decomposition from initial data, and the energy of the reconstructed signals was characterized for analyzing tool wear during the milling process. Then, the filtered cutting force and the cutting vibration features were trained by RBF neural network. Fifteen groups of features were trained by RBF neural network, and three groups of features were used to test RBF neural network. Finally, the results show that the method can accurately monitor the flank wear of milling cutter within a short time, which provides a theoretical basis and experimental scheme for further implementing the on-line tool wear monitoring.

A general information model for the interaction communication between supply chain members is not proposed in the simulation modeling methods of distributed supply chain based on HLA (High Level Architecture) and SCOR (Supply Chain Operation Reference). A single simulation model is used for the modeling of internal structure of federates, by which a clear description of the internal structure and action of supply chain members can not be obtained. In order to solve these problems, a simulation modeling method of distributed supply chain based on HAS (HLA-Agent-SCOR) is put forward, according to characteristics of modeling of HLA, SCOR and Agent. Firstly, the supply chain structure modeling based on HLA that used to build structure model of supply chain is discussed. Secondly, modeling of Agent blocks integrating processes from SCOR and modeling of federates based on Agent are illustrated to create model of supply chain members. Thirdly the simulation framework of distributed supply chain system is designed. Lastly, a simulation example is given and the feasibility of simulation modeling method of distributed supply chain based on HAS is verified.

Traditional electromyography (EMG) pattern recognition did not take into account confounding factors such as electrode shifting, force variation, limb posture, etc., which lead to a great gap between academic research and clinical practice. In this paper, we investigated the robustness of EMG pattern recognition under conditions of electrode shifting, force varying, limb posture changing, and dominant/non-dominant hand switching. In feature extraction, we proposed a method for threshold optimization based on Particle Swarm Optimization (PSO). Compared with the traditional trail & error method, it can largely increase the classification accuracy (CA) by 10.2%. In addition, the hybrid features integrated with discrete Fourier transform (DFT), wavelet transform (WT), and wavelet packet transform (WPT) were proposed, which increased the CA by 30.5%, 25.4%, 22.9%, respectively. We introduced probabilistic neural network (PNN) as a new classifier for EMG pattern recognition, and reported the CA’s obtained by a large variety of features and classifiers. The results showed that the combination of DFT_MAV2 (a novel feature based on DFT) and PNN reached the best CA (45.5%, 14 motions, validated on different hands without re-training).

The intelligent robots have been participating in people’s work increasingly, and challenging to accomplish work autonomously with fully understand human intention through voice interaction. We proposed a robot architecture of hierarchical finite state machine (HFSM) for autonomous mobile manipulator which run on robot operating system (ROS). The system has the abilities to analyze user’s input information, communicate with user to obtain complete intention when the user’s intention is incomplete and transfer the intention to mission plan for executing of tasks. In this paper, we described the operation procedure of system and each component, and designed the experiment scenario to verify the feasibility of the proposed architecture. The experiments result showed our autonomous mobile manipulator achieve the high performance of automation of tasks.

With the development of the wind power industry, wind power has become one of the main green generation energy. At the same time, with the wind power installed capacity increasing, the failure rate gradually growth. As wind turbine is a complex electromechanical equipment, the fault diagnosis for this kind of equipment is also a complicated process. Focused on the current shortage of fault diagnosis knowledge representation, this paper proposes a diagnostic knowledge model for wind turbine and also elaborates the model structure definition with a target to ensure the accuracy of fault diagnosis. Besides, this model can also offer assistance reference model for researchers in related fields to develop advanced methods for sharing and reuse of diagnostic knowledge.

Object tracking is a challenging task especially when occlusion occurs. In this paper, we propose a robust tracking method via structure learning and patch refinement to handle occlusion problem. First, we pose the tracking task as a structured output learning problem to mitigate the gap between pattern classification and the objective of object tracking. Contrary to the random target candidates selection method, we utilize the object independent proposal strategy to generate high quality training and testing samples in structured learning. Second, we over-segment the tracked target to a set of superpixel patches, and then train a background/foreground binary classifier to remove the background patches within the tracked object rectangle area for refining the tracking precision. The objective of target refining is to mitigate tracking model degradation and enhance model robustness for adapting our tracker for long-term and accurate tracking. Experimental results conducted on publicly available tracking dataset demonstrate that the proposed tracking method achieves excellent performance in handling target occlusion.

To allow robots to accomplish manipulation work effectively, one of the critical functions they need is to precisely and robustly recognize the robotic graspable object and the category of the graspable objects, especially in data limited condition. In this paper, we propose a novel multi-loss hierarchical representations learning framework that is capable of recognizing the category of graspable objects in a coarse-to-fine way. Our model consists of two main components, an efficient hierarchical feature learning component that combines kernel features with the deep learning features and a multi-loss function that optimizes the multi-task learning mechanism in a coarse-to-fine way. We demonstrate the power of our proposed system to data of graspable and ungraspable objects. The results show that our system has superior performance than many existing algorithms both in terms of classification accuracy and computation efficiency. Moreover, our system achieves a quite high accuracy (about 82%) in unstructured real-world condition.

Convolutional Neural Networks are powerful tools in object classification which are widely used in Robot Vision. One of the basic requirements of this approach is the demand for a massive data set. However, in many scenarios, it is either economically expensive or difficult (impossible) to collect many valid data with few samples. To this end, in this paper we proposes an automatic approach to collecting data for food industry. First, a robotized data collection system is introduced which uses an industry robot with 6 Degree of Freedoms (DOF). Second, we analysis the key parameters of the proposed system in order to improve the quality of the training model. Finally, the effectiveness of our approach is demonstrated on real experimental platform.

In order to generate the images of the satellites in outer space, a virtual simulation platform was build up based on Visual C++ and OpenGL API. Take the Sinosat-2 Satellite as an example, which is a failure satellite of China. Its projection model was established using computer-graphics technology. Its kinematics formulas were described using quaternions to avoid singularity. A spinning model was set up to demonstrate its motion in 3D space. The platform could simulate the synthetic images of the satellite and then send them to the virtual serve system through TCP/IP in real time.

In this paper, wearable rehabilitation training system for the upper limb based on virtual reality is designed for patients with upper extremity hemiparesis. The six-axis IMU sensor is used to collect the joint training angles of the shoulder and elbow. In view of the patient’s shoulder and elbow joint active rehabilitation training, the virtual rehabilitation training games based on the Unity3D engine are designed to complete different tasks. Its purpose is to increase the interest of rehabilitation training. The data obtained from the experiment showed that the movement ranges of the shoulder and elbow joint reached the required ranges in the rehabilitation training game. The basic function of the system is verified by the experiments, which can provide effective rehabilitation training for patients with upper extremity hemiparesis.

Aiming at improving the low participation and inactive motion intention of patients in traditional gait rehabilitation training, an active gait rehabilitation training system is designed based on Virtual Reality (VR) technology. This project focuses on the design of the gait parameters real-time detecting system and the virtual reality rehabilitation training scene system. Based on an analysis of gait rehabilitation medicine theory, the lower-limb joints range of motion (ROM) and the plantar pressure of the affected limb are selected as the important gait parameters, thus a built-in sensing system is constructed with three inertial measurement units (IMU) and the multi-point force sensing resistors (FSR). Through the wireless Bluetooth communication interface, the lower-limb motor parameters of patients are transmitted into the virtual training games as the motion control signals for character driven in games and the scientific evidence for rehabilitation assessment. Error analysis and compensation method of sampled data are elaborated in this paper. The experiments are carried out about data acquisition, man-machine interaction, and functions of the rehabilitation training scenes. The results show that the active rehabilitation training system is able to assist patients with real-time interaction and immersive sensing and provides better visual feedback information to patients. It improves the training initiative as well as provides an effective means for nerve remodeling.

Aiming at the problem of realtime object detection in humanoid robot, this paper presents a method that combines the Support Vector Machine (SVM) classifier with the Histograms of Oriented Gradients (HOG) feature of local area in the image. In order to reduce time consumption, the paper uses one image segmentation and scan policy to get candidate local area in the image. To evaluate the proposed method, we have established a data set of black and white ball images from RoboCup SPL games. Experimental results have demonstrated that recognition efficiency has been improved greatly and the algorithm can be executed in realtime on the NAO robot that has only a 1.6 GHz CPU.

This paper proposes development of a force leverage mechanism based on the flexure hinges. The primary function of this leverage mechanism is to transform an objective unbalance force/moment to a force sensor in the static unbalance measure system. The measure precision is dependent on the linearity of the force transmission of the force leverage mechanism. The kinematics of the force leverage mechanism is modeled based on the elastic model. The finite element method is used to verify the analytical solutions. Moreover, the effect of the initial external load on the linearity is investigated. Further, the virtual experiment is carried on to verify the linearity and sensitivity. The static unbalance measure system employing the proposed leverage mechanism has the advanced sensitivity of less than 0.03 gcm and performs excellent linearity.

A flapping wing aerial vehicle (FAV) has advantages in the aerodynamic performance, flight flexibility and flight efficiency than the fixed wing and rotary wing flyers. It can be used for military investigation, environment exploration and life rescue. Considering long-distance flight and large payload ratio, we developed a large-sized flapping wing robotic bird. The structure and size comes from that of large-sized birds. The mechanism is first designed by bionics. It has a relatively simple mechanical configuration. Then, the kinematics model is derived to analyze the flapping properties. The sensor system, the control system and the software are also designed according to the requirement of flying control. Then, a prototype was developed. It has a span of 1.1 m, and only weighs 455 g. Finally, a lot of outdoors experiments are performed under nature weather conditions. The continuous flight time is more than 15 min. By analyzing the experimental data, we have found out some interesting phenomena and results.

In this work, a trajectory planning and control system designed for quadrotor unmanned aerial vehicles (UAVs) in automatic field inspection missions is presented. This system controls the UAV by two steps. First, a ground station laptop generates a complete coverage flight path regarding the specific working field. Subsequently, an onboard computer controls the UAV in real-time with position, velocity and acceleration setpoints calculated by a trajectory planning algorithm. The algorithm considers both the dynamic property of quadrotor UAVs and the acceleration and velocity limitations in real working condition. UAV controlled by this system performs a rapid and stable flight in windy environment and can be widely utilised in various domains.

In this paper designs PD-fuzzy sliding mode control for the quadrotor UAV to implement the trajectory-tracking mission. Firstly, dynamic model is introduced for quadrotor UAV. Secondly, the PD-fuzzy sliding mode control is proposed to make the real value of the UAV reach the desired value command, although the UAV is even with system uncertainties and disturbances. The convergence of the complete equations of motion of the UAV is proved by the Lyapunov stability theory. Computer simulation results illustrate the effectiveness of the proposed control schemes.

Bird is one of natural flying masters. It can take advantage of limited energy to fly a long distance. However, the performance of FMAV (Flapping-wing Micro Air Vehicle) is far from the birds. Zoologists have showed that birds can dynamically adjust the flapping amplitude and frequency of its wings. This means that the energy consumption of birds in a variety of flight conditions is minimum. The development of a novel flapping mechanism for FMAV was introduced. The mechanism can achieve independently controllable left and right wings’ flapping-amplitude, both symmetric and asymmetric. The kinematics equations and the static equilibrium equations are deduced for the design and optimization of the AVFM (Amplitude-Variable Flapping Mechanism). The kinetic character of the mechanism is evaluated through computation and simulation. The result shows that the flapping-amplitude of the wings can be changed symmetrically as well as asymmetrically without affecting in-phase flapping motions.

A tail-sitter unmanned aircraft is capable of transition between horizontal and vertical flight. This paper highlights topics of interest to developing a more accurate model to make the simulation results more reasonable. The modeling processes are presented in details. Aerodynamics of this unmanned aircraft is obtained by wind tunnel tests associated with aerodynamic estimation software. Characters of propeller slipstream are analyzed and the mathematical relationships among slipstream velocity, propeller speed, radial location and axial location of propeller plane are deduced from the experiment data. Besides, separate consideration of the propeller slipstream on wings and control surfaces gives better estimations on the dynamic pressure. Models of actuators and motors are also obtained through some tests to make the results reliable. Furthermore, a simple controller is designed to implement the hover attitude control.

Flapping-wing Micro Aerial Vehicle (FWMAV) is a kind of aircraft which can produce lift and thrust by flapping the wings just like birds or insects. Wind tunnel experiment played an important role in the Micro Aerial Vehicle (MAV) research. This paper introduces wind tunnel experiments of a pair of flexible flapping wings, and analyzes mean and instantaneous characteristics of aerodynamic force and torque. According to the characters of force and torque, they are divided into average signals and periodic alternating signals. Average signals are related to motion characteristics of aircraft, like fixed wing aerial vehicle. Meanwhile, periodic alternating signals are related to periodic flapping up and down, and the frequency is the same as flapping frequency. Finally, the article establishes an available longitudinal flight dynamic model of FWMAV using small perturbation theory to deal with average signals and the model was verified by the real flight tests.

The subsurface access technology is of great significance to study evolution of stars and probe landing, and the low velocity robotic penetrator (LVRP) is a commonly used detection technique. In this study, a novel LVRP based on ampere force (AFRP) is proposed, which uses the ampere force of the coils located in high intensity magnetic field as the driving force. The AFRP is composed of stator and mover. In the working state, the mover is subjected to a linear reciprocating motion, colliding with the stator, and displacement occurs. Firstly, parametric modeling of AFRP is carried out and the internal magnetic field distribution is analyzed. The results show that the magnetic induction between the two poles can be increased by the relative mounting of the permanent magnets, and the magnetic induction is affected by the gap and the diameter of the permanent magnets. Then, ISIGHT® software is used for integrated optimization, whose purpose is to optimize the design variables and make the output force of each unit reaches the maximum when the casing size of the AFRP is certain. Furthermore, to achieve the optimization design goal that the overall output force is maximum. According to the optimization results, the prototype was developed. The experimental results show that the AFRP operating frequency can be controlled in the range of 1 ~ 10 Hz, the average displacement of each motion cycle is 2 mm. The proposed AFRP can output a larger driving force by changing the current and the number of units. In addition, the AFRP can be used as a novel actuator for robots.

A flexible manipulator with concentrated mass at its end is simplified to be a Bernoulli-Euler beam module in this paper. The oscillatory differential equation and system decoupling equation of the one DOF flexible manipulator is deduced according to the module simplified from practical engineering. And then, based on the establishment of dynamic model, we get the state of space expression through the Lagrange equation. Take the first order modal parameters of the flexible manipulator to track the angular displacement of the manipulator and the deflection of its end. According to the established state space expression, a comparison is done between the PD controller and the integral separation PID controller in the active control of flexible manipulator through simulation in Simulink. Result shows that the integral separation PID controller has obvious advantage than the PD controller in the control of flexible manipulator. The former is better to quickly arrive at the specified position and both its static error and overshoot are smaller.

Space debris has increased with recent launch missions greatly. Former active debris removal tests using space robot mainly focused on the fundamental technology of target recognition, motion control and path planning. However, robot contacts directly with the surface of targets with large mass and angular momentum will cause severe collision problems. A target de-tumbling strategy is proposed in this paper by using two manipulators. Each arm is equipped with magnetic coil, which can generate eddy current in conductive targets and gradually de-tumble rotation without contact. The three-dimension rotation model of a discarded satellite and upstage is established based on its distribution of the moment of inertia and the safe working space of the robots is calculated. By analyzing the point of application and direction of the magnetic force, an optimized de-tumbling trajectory for the robot is presented to minimize the de-tumbling time by reducing the targets’ angular momentum. At last, a simulation is processed to verify the optimized de-tumbling method.

Due to a large number of electrical cables in the joints of traditional space manipulator, the assembly, testing and maintenance are very inconvenient. In this paper, a modular joint with light weight and wireless communication link is developed to solve the above problem. The performance of the joint is firstly determined according to task requirement. Then, the mechanical structure is designed to meet the performance requirement. It is mainly constructed by a brushless DC motor, harmonic reducer, bearing, and supporting structure. The electrical sub-system includes sensors, servo controller and WIFI communication link. Multiple sensors, including three hall sensors, an incremental magnetic encoder, an absolute magnetic encoder, and a torque sensor are installed in the joint. The vector control is used for motor. The WIFI communication link is designed for the communication between the central controller and the servo controller. Therefore, the task command can be sent through the WIFI link. Electrical cables between the central controller and joint servo controller are not required. Finally, the developed joint is tested on a test platform. The experimental results verify the performance of the joint.

Tethered Space Net (TSN) has been proposed since there is an increasing threat of space debris to spacecraft and astronauts in recent years. In this paper, we propose an improved TSN, named Maneuverable-Net Space Robot (MNSR), which has four maneuvering units in its four corners (square net). The four maneuverable units make the MNSR controllable. Because of autonomous maneuverability, the attitude dynamics of the platform, master tether and flexible net are strongly coupled. In order to design an effective controller to maintain the configuration of the maneuverable net, an accurate dynamics model of MNSR based on the Lagrangian method is derived. In our model, we consider the three-dimensional attitude of the platform, master tether and maneuvering-net as well. Due to the vibration of the in-plane and out-of-plane angles of the net tethers, feedback control is employed for MNSR. The simulation results demonstrate that the proposed control is efficient and suitable for the MNSR system.

In this paper, space manipulator man machine cooperation system structure and operation mode of Mir Space Station and International Space Station are reviewed briefly. Based on this, system structure, control mode, operation mode of Chinese first space manipulator and dexterous hand man machine cooperation experiment are introduced. Mathematics models and ground tests methodology of man machine cooperation are given. Via on orbit experiment, flight scheme and ground validation of space manipulator and dexterous hand man machine cooperation are reasonable. Through Chinese first on orbit space manipulator and dexterous hand man machine cooperation technology experiments, not only the technology foundation of man machine cooperation is established, but also it is accumulated experience for Chinese space robot.

In this paper, an inverse kinematics analysis algorithm of a robot is proposed based on the analysis of the configuration and kinematics of the robot. With the inverse kinematics method, the working path of the robot’s joints is planned. Based on multi-body dynamics, finite element analysis, topology optimization and PID control, using co-simulation, this paper presents the method including Structural Optimization, Mechanical System Optimization and Controller Optimization. The main bearing part (large arm) is selected as the optimization object, in order to improve the repeated positioning accuracy of the robotic robot and verify the collaborative optimization methods, the effectiveness of the method is improved by comparing the error of the end position with the same working path by comparing the robotic system before and after the collaborative optimization.

Using tethered space robot (TSR) for active debris removal (ADR) is promising but subject to collision and entanglement due to the debris tumbling. To detumble the towed debris, this paper proposes the nonlinear model predictive control (NMPC) based coordinated control strategy. The TSR consists of a gripper for capture, thrusters and a tethered manipulator (TM) with variable length to which the tether is attached. The proposed strategy works in the way that the TM coordinates with the thrusters for de-tumbling by changing its length accordingly so that the tension torque can be adjusted. The attitude model of the debris is first established, followed by the definition of attitude equilibrium. The NMPC is then designed with the prediction model discretized by 4-order Runge-Kutta method. Simulation results validate this strategy and show that the debris attitude can maneuver to the equilibrium smoothly in the presence of the constraints on TM and thrusts.

This paper achieves the design of the key component of space manipulator – the telescopic boom and develops the principle prototype, to solve the contradiction between large working space, small load and fine operation, high rigidity, high precision in space manipulator handling operations. The stiffness test platform is developed for the telescopic boom, and through the test the boom meets the design requirements. The control strategy used to stretch the boom is also discussed in detail. Relevant experiments show that the telescopic boom can successfully achieve the telescopic movement.

A typical servicing operation in space mainly includes three phases: capturing the target, re-orientating of the whole system with the target, and repairing the target. A method is proposed to achieve re-orientating the space robot system, planning the manipulator configuration and spacecraft orientation at the same time. Firstly, angular momentum preloaded in the manipulator will provide a favorable condition for the capturing and reorientation after capturing, so the manipulator moves with some initial velocity after capturing. The constraints on the manipulator and the objective function are defined according to the planning problem. Then the joint trajectories are parameterized by using sinusoidal polynomials functions and the cost function is proposed according to the accuracy requirements. Finally, Particle Swarm Optimization (PSO) is used to search for the global optimal resolution of the parameters. When the parameters are found, each joint trajectory can be determined. The simulation results show that this method is better than other approaches both in convergence rate and accuracy and the joints trajectories are very smooth and suitable for control.

This paper presents an inverse kinematics solution method for a 7-degrees-of-freedom (DOF) redundant robot with offset shoulder and wrist joints. Aiming at the plan-points of the 7-DOF redundant robot, we propose an iterative method to solve the inverse kinematics for a specific application scenario, that is, the orientation matrices and arm angle keep the same values during the movement of robot which can be found in some specific operations, like painting on a plane. The position, orientation and arm angle of current plan-point are used to calculate the corresponding values of next plan-point. Through iterative calculation, the inverse kinematics solutions can be obtained with the characteristic that the orientation matrices and arm angle maintain the same values for all plan-points. The comparison and analytical results show this iterative solving method is effective.

This work proposes a kinematic modeling and a kinematic nonlinear controller for an autonomous aerial mobile manipulator robot that generates saturated reference velocity commands for accurate waypoint path following. The dynamic compensation controller is considered through of a helicopter-inner-loop system to independently track four velocity commands: forward, lateral, up/downward, and heading rate; and arm-inner-loop system to independently track angular velocity commands. Stability and robustness of the complete control system are proved through the Lyapunov method. Finally, simulation results are presented and discussed, which validate the proposed controller.

Aiming at eliminating vibration generated during the motion state switch of robotic joints, this study proposes a magnetorheological fluids (MRFs) based soft actuator to achieve semi-active vibration control. In this paper, the configuration of the MRFs actuator is described firstly, followed by the theoretical modeling of the magnetic circuit and the transmitted torque. Then, the structural model of the actuator is designed and presented. After these, the influences of working induction and speed difference on both total transmitted torque and controllable coefficient are numerically calculated. Finally, an electromagnetic simulation is carried out with ANSYS 10.0® to verify the designed magnetic circuit of the actuator. The results indicate that the working induction holds a strong impact on both total transmitted torque and controllable coefficient; however, the influences of speed difference were relatively slight. Moreover, the designed circuit is proved to fulfill the requirements of both induction intensity and uniformity.

In this paper, a proportional-derivative (PD) controller combined with a disturbance observer is adopted for a designed magnet-driven nano positioning stage. Firstly, the magnet-driven positioning stage with long stroke and nano positioning accuracy is introduced in detail. The stage adopts air bearings to eliminate complicated nonlinear friction effect. The actuator of the stage is a permanent magnet synchronous linear motor which uses ironless windings to eliminate the cogging force and attenuate the reluctance force. Then, a PD controller combined with a disturbance observer is used to control the stage. The validation experiment is carried out based on Matlab/Simulink Real-Time toolbox. The experimental results shows that the maximum travel range is 50 mm and positioning accuracy no bigger than 3 nm. The compared experiment with a conventional proportional-integral-derivative controller shows ten times better position accuracy is acquired by using proposed controller.

The precision positioning technology has developed for a long time and played a significant role in many fields. However, most precision positioning systems designed before only have one or two degrees of freedom, which greatly limits the application. In view of this, a kind of three-degrees-of-freedom (3-DOF) macro-micro precision positioning system is investigated and analyzed in the paper. The macro-micro precision positioning system is designed suitable for application in vacuum environment, which includes two main parts such as 3PRR (3 degrees of freedom, each branch consists of a prismatic pair (P) drive and two rotating pairs (R)) planar parallel platform and piezoelectric micro stage. Before conducting the experiment of macro-micro precision positioning system, it is necessary to investigate the loading effect on the 3PRR planar parallel platform. The loading experiment has showed that different loads have some effect on positioning accuracy and standard variance, but the positioning error is much less than the travel range of micro positioner. Then the experiment combining 3PRR planar parallel platform and piezoelectric micro stage has been carefully conducted by using laser interferometer as feedback control measurement. The experimental results demonstrate that the requirement of high positioning precision and large travel range can be simultaneously met by using the macro-micro precision positioning system.

At present spacer bar cable clamp used for high voltage transmission line has been assembled manually at home, which lead to low production efficiency and unstable product quality. In order to achieve the high quality requirement of spacer bar cable clamp, a coordinated control system for spacer bar assembly equipments based on S7-200 PLC was developed. The composition of the control system was determined according to spacer bar assembly process and the hardware and the software included in the control system were designed. The new automatic control system greatly could improve the assembly efficiency and product quality, substituted for the hand labor.

To realize super-resolution optical and terahertz imaging by the sub-wavelength hole arrays with extraordinary optical transmission (EOT) performance, the related sampling template structure is studied. In this paper, Hadamard matrix, cyclic S matrix and matrix chosen by random matrix selector are defined as a small probability matrix. The sub-wavelength coding imaging template based on the structure of this small probability matrix have extraordinary optical transmission (EOT) performance, which is the basis for achieving super-resolution imaging. It is showed that, in the case of array detector, high signal-to-noise ratio (SNR) improvement (exceeded 22) can be obtained only by sampling one frame using the imaging template designed by 3 order cyclic S matrix. For single detector, the SNR improvement using various order of cyclic S matrix designed imaging template is higher than N a quadratic root obtained by Hadamard matrix in classical Hadamard transform optical imaging method, and much higher than half a quadratic root of N by classical cyclic S matrix.

The forming of balloons used in medical treatment is a kind of “black box art”. When a new balloon is being developed, the process parameters and tube dimension are usually determined by a method of trial and error. This method is inefficient in current rapid development of computer technology. Numerical simulation is expected to replace the experiments and experience to guide the development of the new products. In this study, the moulding of the balloon was simulated by a finite element method and the results obtained from the simulation agreed with that of the experiments under the same actual process parameters. Therefore the numerical simulation used is feasible for the process of balloons forming. The effect of process parameters on the wall thickness of balloon was analyzed based on orthogonal design method. The results showed that the effect of first stretch rate on the wall thickness of the balloon was the most significant compared with other process parameters. A regression model of the relationship between wall thickness and the process parameters was established, which could be used to guide the selection of production process parameters.

This paper presents a novel geometric simulation technique for multi-axis Additive Manufacturing (AM). In the proposed methodology, additive swept volume elements, which are represented with Tri-dexel models, are formulated for updating the virtual additive material. Triangular meshes are extracted from the Tri-dexel models using Marching Cube Algorithm for visualization. With the proposed methodology, either three-axis or five-axis AM tool paths of sculpture surfaces can be simulated before actually manufactured. Workpieces after additive manufacturing can be used as blanks for further Subtractive Manufacturing(SM). We developed a geometric simulation software based on the multi-axis AM simulation algorithm, and we carried out an actual three-axis AM experiment to compare with the simulation results. Computer implementation and practical example demonstrate the feasibility of the proposed multi-axis AM simulation method.

Raw rubber or rubber items are vulnerable to heat, oxygen, light and other factors during the processing, storing or using, due to exposure to the natural environment or a particular working environment, and easy to undergo physical or chemical change, such as the softening, sticky for crude rubber, and cracking, mildewy and brittle for rubber products which would degrade the material properties or even make it unusable. The loss of property caused by such rubber aging is up to hundreds of millions of dollars annually. Therefore how to accurately predict the rubber aging life and how to choose the appropriate test method are especially important for the selection of suitable rubber and reduction of the cost of existing aging experiment. Based on the summarizing and comparison of various kinds of rubber fatigue aging theories and experiment methods, a predicting rubber aging life software with VC++ programming language was developed. The software structure mainly included the theoretical prediction module, the commonly used rubber material data module, user guidance module. It could be used to predict rubber fatigue life, optimize the product material and help the manufacturing process more efficient implementation.

In the mass production assembly process, the fixture system’s reliability is vital for products’ quality. The failure of the fixture system depends not only on the assembly operation times, but also upon the system degradation which is caused by the original manufacturing accuracy of fixtures, assembly forces, qualities of subassemblies from upstream assembly stations etc. In this paper, we propose a dynamic preventive maintenance policy for the fixture components based on the system reliability model of a multi-station assembly process. The proposed reliability model not only considers the degrading fixture components and other factors in the station, but also the dependence of process factors across the stations. Based on the reliability model and a given cost, an optimization method for the manufacturing tolerances and a dynamic maintenance schedule of locating pins are presented. At last, a body side assembly case is given to illustrate the proposed method.

The complex surfaces are usually approximated as Z-maps, point based models or facet models to simplify the computation for cutter positions. However, such surface approximation error may result in the conflict between the accuracy of cutter positions and the computational efficiency, and C1 discontinuity for tool paths and tool orientations. In this paper, an efficient cutter-freeform surfaces projection (CFSP) method based on the cutter-facet model projection (CFMP) method is proposed for five-axis tool path computation to solve the above issues. Motivated by the Newton-Raphson algorithm, the tangent plane-based search method is developed to find the accurate cutter contact points on surfaces. The golden section-based search algorithm is further carried out to deal with the convergence oscillation problem. Simulation on the part models validates that the algorithm is effective to improve the machining accuracy.

In this paper, a method to optimize the milling process parameters based on the real-coded self-adaptive genetic algorithm (RAGA) and Grey relational analysis (GRA) is proposed. Experiments have been designed with four input milling process parameters at four different levels. The RAGA coupled with GRA has been applied for solving the proposed optimization problem to achieve the desired machined surface quality characteristics. Simulation experiments give the optimal parametric combination. Furthermore, experiments for the machined surface topography with the initial and optimal combination of milling process parameters are implemented and the results verify the feasibility of the proposed method.

In this paper, we demonstrate a Hybrid Manufacturing software-HybridCAM. HybridCAM provides a variety of tool path generation methods for sculpture surfaces. The tool path patterns include contour-parallel pattern, zigzag pattern, and helical pattern. HybridCAM can input multiple format of models, such as IGS, STL, and STP, and can be applied to multi-axis Additive Manufacturing (AM) and multi-axis Subtractive Manufacturing (SM). We present several tool path patterns for different parts and different applications generated by HybridCAM. An actual experiment was carried out to verity the feasibility and practicality of the software.