Towards Autonomous Robotic Systems

Myoelectric hand prostheses have the potential to improve the quality of life of hand amputees. Still, the rejection rate of functional prostheses in the adult population is high. One of the causes is the long time for fitting the prosthesis and the lack of feedback during training. Moreover, prosthesis control is often unnatural and requires mental effort during the training. Virtual and augmented reality devices can help to improve these difficulties and reduce phantom limb pain. Amputees can start training the residual limb muscles with a weightless virtual hand earlier than possible with a real prosthesis. When activating the muscles related to a specific grasp, the subjects receive a visual feedback from the virtual hand. To the best of our knowledge, this work presents one of the first portable augmented reality environment for transradial amputees that combines two devices available on the market: the Microsoft HoloLens and the Thalmic labs Myo. In the augmented environment, rendered by the HoloLens, the user can control a virtual hand with surface electromyography. By using the virtual hand, the user can move objects in augmented reality and train to activate the right muscles for each movement through visual feedback. The environment presented represents a resource for rehabilitation and for scientists. It helps hand amputees to train using prosthetic hands right after the surgery. Scientists can use the environment to develop real time control experiments, without the logistical disadvantages related to dealing with a real prosthetic hand but with the advantages of a realistic visual feedback.

This study explores how new robotic systems can help library users efficiently locate the book they require. A survey conducted among Imperial College students has shown an absence of a time-efficient and organised method to find the books they are looking for in the college library. The solution implemented, LibRob, is an automated assistive robot that gives guidance to the users in finding the book they are searching for in an interactive manner to deliver a more satisfactory experience. LibRob is able to process a search request either by speech or by text and return a list of relevant books by author, subject or title. Once the user selects the book of interest, LibRob guides them to the shelf containing the book, then returns to its base station on completion. Experimental results demonstrate that the robot reduces the time necessary to find a book by 47.4%, and left 80% of the users satisfied with their experience, proving that human-robot interactions can greatly improve the efficiency of basic activities within a library environment.

The development of robotic-assisted extracorporeal ultrasound systems has a long history and a number of projects have been proposed since the 1990s focusing on different technical aspects. These aim to resolve the deficiencies of on-site manual manipulation of hand-held ultrasound probes. This paper presents the recent ongoing developments of a series of bespoke robotic systems, including both single-arm and dual-arm versions, for a project known as intelligent Fetal Imaging and Diagnosis (iFIND). After a brief review of the development history of the extracorporeal ultrasound robotic system used for fetal and abdominal examinations, the specific aim of the iFIND robots, the design evolution, the implementation details of each version, and the initial clinical feedback of the iFIND robot series are presented. Based on the preliminary testing of these newly-proposed robots on 42 volunteers, the successful and reliable working of the mechatronic systems were validated. Analysis of a participant questionnaire indicates a comfortable scanning experience for the volunteers and a good acceptance rate to being scanned by the robots.

People with temporary or permanent disabilities often struggle to perform daily tasks, and this loss of agency can cause undue emotional distress. This paper presents “Eduardo” (Fig. 1); a low cost prototype of an assistive robot arm featuring a passive compliant element in the end effector of the arm, to simplify the control of the arm, and ensure the safety of the user during physical interaction. As a demonstration of the prototype, the arm was programmed to comb hair. The robot is envisioned as an open source platform for research into the use of compliance for assistive robots. Fig. 1. The Combing robot

Developing autonomous assistants to help with domestic tasks is a vital topic in robotics research. Among these tasks, garment folding is one of them that is still far from being achieved mainly due to the large number of possible configurations that a crumpled piece of clothing may exhibit. Research has been done on either estimating the pose of the garment as a whole or detecting the landmarks for grasping separately. However, such works constrain the capability of the robots to perceive the states of the garment by limiting the representations for one single task. In this paper, we propose a novel end-to-end deep learning model named GarmNet that is able to simultaneously localize the garment and detect landmarks for grasping. The localization of the garment represents the global information for recognising the category of the garment, whereas the detection of landmarks can facilitate subsequent grasping actions. We train and evaluate our proposed GarmNet model using the CloPeMa Garment dataset that contains 3,330 images of different garment types in different poses. The experiments show that the inclusion of landmark detection (GarmNet-B) can largely improve the garment localization, with an error rate of 24.7% lower. Solutions as ours are important for robotics applications, as these offer scalable to many classes, memory and processing efficient solutions.

In the last decade, soft robots demonstrated their distinctive advantages compared to ‘hard’ robots. Soft structures can achieve high dexterity and compliance. However, only low forces can be exerted, and more complicated control strategies are needed. Variable stiffness robots offer an alternative solution to compensate for the downsides of flexible robots. One of the most common approach in the development of variable stiffness robots is the use of granular jamming. In this paper a variable stiffness manipulator based on granular jamming is studied. Here, we propose the use of soft and deformable spherical particles instead of commonly used rigid particles. Further on, we evaluate the performance of the soft particles under vacuum. In addition, a comparison between our approach and the standard approach to granular jamming is presented. The proposed soft particles show good performance in terms of their capability of compacting and squeezing against each other to achieve a high-stiffness robot arm.

OctoMap is a popular 3D mapping framework which can model the data consistently and keep the 3D models compact with the octree. However, the occupancy map derived by OctoMap can be incorrect when the input point clouds are with noisy measurements. Point cloud filters can reduce the noisy data, but it is unreasonable to apply filters in a sparse point cloud. In this paper, we present a k-nearest neighbours (k-NN) based inverse sensor model for occupancy mapping. This method represents the occupancy information of one point with the average distance from the point to its k-NN in the point cloud. The average distances derived by all the points and their corresponding k-NN are assumed to be normally distributed. Our inverse sensor model is presented based on this normal distribution. The proposed approach is able to deal with sparse and noisy point clouds. We implement the model in the OctoMap to carry out experiments in the real environment. The experimental results show that the 3D occupancy map generated by our approach is more reliable than that generated by the inverse sensor model in OctoMap.

This paper presents an elastomer-based tactile sensor that can sense the tactile information in the form of pressure distribution. Our proposed sensor uses a piece of coated elastomer with thin conical pins underneath as the touch medium. The elastomer consists of 91 pins arranged in a honeycomb pattern, each pin can be regarded as a tactile sensing element. They are spaced at 1.5 mm in x and y direction. Each tactile element transfers the applied pressure value into a circular image pattern which can be captured by a camera placed at the end of the sensor structure. The applied pressure over the sensing array can be computed by processing the area of each sensing element. MATLAB is used to process the received images relating the applied pressure to the activated pixels in each circular pattern of the tactile element, and further visualizing the pressure distribution on a reconstructed surface of the sensor. This paper presents the development principle and fabrication process of the proposed sensor. The experimental results have proven the viability of the sensing concept; the prototype sensor can effectively detect single-point touch caused by objects with different dimensions and multi-point touch interactions with a spacing of more than 2.5 mm.

Soft sensors are crucial to enable feedback in soft robots. Soft capacitive sensing is a reliable technology that can be embedded into soft pneumatic robots for obtaining proprioceptive and exteroceptive feedback. In this paper, we model a soft capacitive sensor that measures both the actuated state as well as applied external forces. We develop a Finite Element Model using a multiphysics software (COMSOL®). Using this model, we investigate the change in capacitance with the application of external force, for a range of different internal pressures and strains. We hope this study is helpful in understanding the coupling of internal inputs and external stimuli on the feedback obtained from the sensors and help us design better sensory systems for soft robots.

This work intends to contribute to transfer human in-hand manipulation skills to a dexterous prosthetic hand. We proposed a probabilistic framework for both human skill representation and high efficient recognition. Gaussian Mixture Model (GMM) as a probabilistic model, is highly applicable in clustering, data fitting and classification. The human in-hand motions were perceived by a wearable data glove, CyberGlove, the motion trajectory data proposed and represented by GMMs. Firstly, only a certain amount of motion data were used for batch learning the parameters of GMMs. Then, the newly coming data of human motions will help to update the parameters of the GMMs without observation of the historical training data, through our proposed incremental parameter estimation framework. Recognition in the research takes full advantages of the probabilistic model, when the GMMs were trained, the log-likelihood of a candidate trajectory can be used as a measurement to achieve human in-hand manipulation skill recognition. The recognition results of the online trained GMMs show a steady increase in accuracy, which proved that the incremental learning process improved the performance of human in-hand manipulation skill recognition.

This paper presents recent work on the subject of measurement variance in Ultra-Wide Band localisation systems. Recent studies have shown the utility in more rigorous noise characterisation of sensor inputs used in state estimation systems such as the Extended Kalman Filter. This investigation strategy is extended to using data collected during trials of such state estimation algorithms using an Unmanned Ground Vehicle, for the generation of variance maps of the testing environments. The feasibility of building variance models from this data is discussed, and other applications for the information is proposed. As there exist circumstances where the practice of moving the agent around a space incrementally is not practicable, such as in the case of Unmanned Aerial Vehicles, or in restricted spaces, an alternate method is needed. From the results it can be concluded that the use of data collected during standard operation in the environment is not sufficient for initial characterisation of localisation sensors. Initial analysis of this data was also utilised to investigate the effects of environmental factors.

This paper presents MoDSeM, a software framework for cooperative perception supporting teams of robots. MoDSeM aims to provide a flexible semantic mapping framework able to represent all spatial information perceived in missions involving teams of robots, and to formalize the development of perception software, promoting the implementation of reusable modules that can fit varied team constitutions. We provide an overview of MoDSeM, and describe how it can be applied to multi-robot systems, discussing several sub-problems such as history and memory, or centralized vs distributed perception. Aiming to demonstrate the functionality of our prototype, preliminary experiments took place in simulation, using a $$100 \times 100 \times 100$$ 100 × 100 × 100 m simulated map to demonstrate its ability to receive, store and retrieve information stored in semantic voxel grids, using ROS as a transport layer and OpenVDB as a grid storage mechanism. Results show the appropriateness of ROS and OpenVDB as a back-end for supporting the prototype, achieving a promising performance in all aspects of the task. Future developments will make use of these results to apply MoDSeM in realistic scenarios, including multi-robot indoor surveillance and precision forestry operations.

This paper investigates the application of temporal planning to multiple robots in long-term missions, using the OPTIC and POPF temporal planners. We design a new planning domain, motivated by a realistic indoor-outdoor scenario. In particular, we investigate plan concurrency, makespan and plan generation time in the multi-robot problem and propose a schema which has been shown to improve plan quality while significantly reducing planning time for the multi-agent problem. Experiments are done in simulation using ROS and Gazebo, and demonstrated in missions with concurrent actions. The ROSPlan framework is also extended to work with multiple robots and used to integrate the planners in ROS. OPTIC provides the best overall solution considering the domain complexity and mission execution in the environment.

The study of combined task and path planning has mainly focused on feasibility planning for high-dimensional, complex manipulation problems. Yet the integration of symbolic reasoning capabilities with geometric knowledge can address optimal planning in lower dimensional problems. This paper presents a dynamic, anytime task and path planning approach that enables mobile robots to autonomously adapt to changes in the environment. The planner consists of a path planning layer that adopts a multi-tree extension of the optimal Transition-based Rapidly-Exploring Random Tree algorithm to simultaneously find optimal paths for all movement actions. The corresponding path costs, derived from a cost space function, are incorporated into the symbolic representation of the problem to guide the task planning layer. Anytime planning provides continuous path quality improvements, which subsequently updates the high-level plan. Geometric knowledge of the environment is preserved to efficiently re-plan both at the task and path planning level. The planner is evaluated against existing methods for static planning problems, showing that it is able to find higher quality plans without compromising planning time. Simulated deployment of the planner in a partially-known environment demonstrates the effectiveness of the dynamic, anytime components.

In this work, we present the development of a neuro-inspired approach for characterizing sensorimotor relations in robotic systems. The proposed method has self-organizing and associative properties that enable it to autonomously obtain these relations without any prior knowledge of either the motor (e.g. mechanical structure) or perceptual (e.g. sensor calibration) models. Self-organizing topographic properties are used to build both sensory and motor maps, then the associative properties rule the stability and accuracy of the emerging connections between these maps. Compared to previous works, our method introduces a new varying density self-organizing map (VDSOM) that controls the concentration of nodes in regions with large transformation errors without affecting much the computational time. A distortion metric is measured to achieve a self-tuning sensorimotor model that adapts to changes in either motor or sensory models. The obtained sensorimotor maps prove to have less error than conventional self-organizing methods and potential for further development.

Inspection is a hot topic of robotics recently, and there are many different ways to solve the inspection problem. In this paper, we propose a new framework for a robust and efficient inspection of the entire workspace in a watchman route based on automatically generated waypoints. The framework architecture design includes several relevant technologies and refines algorithms such as medial axis transformation, shortest path approximation, and Monte-Carlo search for finding tours. This framework is evaluated in a client-server system: the simulation of the robot is run on Unity, while data processing is executed in a Python server. Experimenting with this approach, the measured inspection coverage of the workspace on random terrains was at least 99.6%.

Assisting people with daily living tasks in their own homes with a robot requires a navigation through a cluttered and varying environment. Sometimes the only possible path would be blocked by an obstacle which needs to be moved away but not into other obstructing regions like the space required for opening a door. This paper presents semantic assisted path planning in which a gridded semantic map is used to improve navigation among movable obstacles (NAMO) and partially plan simple household tasks like cleaning a carpet or moving objects to another location. Semantic planning allows the execution of tasks expressed in human-like form instead of mathematical concepts like coordinates. In our numerical experiments, spatial planning was completed well within a typical human-human dialogue response time, allowing for an immediate response by the robot.

In the coming years, incremental automation will be the main challenge in the development of highly versatile helicopter technologies. To support this effort, vision-based systems are becoming a mandatory technological foundation for helicopter avionics. Among the different advantages that computer vision can provide for flight assistance, navigation in a GPS-denied environment is an important focus for Airbus because it is relevant for various types of missions. The present position paper introduces the different available SLAM algorithms, along with their limitations and advantages, for addressing vision-based navigation problems for helicopters. The reasons why Visual SLAM is of interest for our application are detailed. For an embedded application for helicopters, it is necessary to robustify the VSLAM algorithm with a special focus on the data model to be exchanged with the autopilot. Finally, we discuss future decisional architecture principles from the perspective of making vision-based navigation the 4th contributing agent in a wider distributed intelligence system composed of the autopilot, the flight management system and the crew.

Research in swarm robotics has shown that robot swarms are effective in the exploration of unknown environments. However, little work has been devoted to port the exploration capabilities of robot swarms into the context of mapping. Indeed, conceiving robot swarms that can map an unknown environment in a robust, scalable, and flexible way is an open issue. In this paper, we investigate a swarm mapping method in which robots first individually map the environment by random walk and then, we merge their maps into a single, global one. We focus on five variants of random walk and we compare the quality of the maps that a swarm produces when exploring the environment using these variants. Our experiments with ten e-puck robots show that, despite the individual maps being incomplete by themselves, it is possible to collectively map the environment by merging them. We found that the quality of the map depends on the exploration behavior of the individuals. Our results suggest that one of the variants of random walk, the ballistic motion, gives better mapping results for closed environments.

With the continued advancement of autonomous vehicles and their implementation in public roads, accurate detection of vulnerable road users (VRUs) is vital for ensuring safety. To provide higher levels of safety for these VRUs, an effective detection system should be employed that can correctly identify VRUs in all types of environments (e.g. VRU appearance, crowded scenes) and conditions (e.g. fog, rain, night-time). This paper presents optimal methods of sensor fusion for pedestrian and cyclist detection using Deep Neural Networks (DNNs) for higher levels of feature abstraction. Typically, visible sensors have been utilized for this purpose. Recently, thermal sensors system or combination of visual and thermal sensors have been employed for pedestrian detection with advanced detection algorithm. DNNs have provided promising results for improving the accuracy of pedestrian and cyclist detection. This is because they are able to extract features at higher levels than typical hand-crafted detectors. Previous studies have shown that amongst the several sensor fusion techniques that exist, Halfway Fusion has provided the best results in terms of accuracy and robustness. Although sensor fusion and DNN implementation have been used for pedestrian detection, there is considerably less research undertaken for cyclist detection.

In environments featuring hazards (e.g., debris, holes in the ground), robot navigation can be challenging. Robot’s sensors alone might be not able to guarantee timely detection of the threats. In such situations, the presence of nearby humans can be exploited to support safe robot navigation. The human can proactively provide advisory information and issue warnings. Unfortunately, verbally expressed human’s inputs are usually quite imprecise or ambiguous when referring to spatial elements. We consider how to model the inherently imprecise and sporadic “human sensor” by using the formalism of imprecise probabilities, and how to use the model to build maps fusing robot sensor data and human inputs. Map information is used for path planning, searching for paths that balance survivability and efficiency (e.g., time). In a number of simulation scenarios we study the effectiveness of our approach compared to standard ways to build the map and perform path planning.

Space robots are key to the establishment of a new era of low-cost in-orbit operations. Given the complexities involved in designing and operating of a space robot, several challenges arise and developing new advanced methodologies for control and motion planning is essential. Finding an optimal trajectory for the space robot to attain an out-of-reach grasping point on the target or when the motion of the arm is restricted by singular configurations or obstacles, is a difficult task using the Degrees of Freedom (DoF) of the arm only. Hence, using the redundancy offered by the extra degrees of freedom of the spacecraft base to help the arm reach the target whilst avoiding singularities and obstacles is mission critical. In this paper, an optimal path planning algorithm using Genetic Algorithm was developed for a controlled-floating space robot that takes advantage of the controlled motion of the spacecraft base to safely reach the grasping point. This algorithm minimises several cost functions whilst satisfying constraints on the velocity. Moreover, the algorithm requires only the Cartesian location of the grasping point, to generate a path for the space robot without a priori knowledge of any desired path. The optimal trajectory is tracked using a nonlinear adaptive $$H_{\infty }$$ H ∞ controller for the simultaneous motion of both the manipulator and the base spacecraft. The results presented prove the efficacy of the path planner and controller and it is based on a six DoF manipulator mounted to a a six DoF spacecraft base.

The performance of a multi-robot team varies when certain environmental parameters change. The study presented here examines the performance of four task allocation mechanisms, compared across a mission landscape that is defined by a set of environmental conditions. The landscape is categorised by three dimensions: (1) single-robot versus multi-robot tasks; (2) independent versus constrained task correspondence; and (3) static versus dynamic allocation of tasks with respect to mission execution. Two different task scenarios and two different starting formations were implemented with each environmental condition. Experiments were conducted on teams of simulated and physical robots, to demonstrate the portability of the results. This paper investigates the “static allocation” portion of the mission landscape, filling in a gap that has not been investigated previously. Experimental results are presented which confirm that the previous conclusion still holds: there is no single task allocation mechanism that consistently ranks best in performance when tasks are executed.

We present a vision-based motion analysis method for single and multiple mobile robots which allows quantifying the robot’s behaviour. The method defines how often and for how much each of the robots turn and move straight. The motion analysis relies on the robot trajectories acquired online or offline by an external camera and the algorithm is based on iteratively performed a linear regression to detect straight and curved paths for each robot. The method is experimentally validated with the indoor mobile robotic system. Potential applications include remote robot inspection, rescue robotics and multi-robotic system coordination.

Despite various existing works on intuitive human-robot interaction (HRI) for teleoperation of robotic manipulators, to the best of our knowledge, the following research question has not been investigated yet: Can we have a teleoperated robotic manipulator that simply copies a human operator’s bare hand posture and gesture in a real-time manner without having any hand-held devices? This paper presents a novel teleoperation system that attempts to address this question. Firstly, we detail how to set up the system practically by using a Universal Robots UR5, a Robotiq 3-finger gripper, and a Leap Motion based on Unity and ROS, and describe specifically what information is communicated between each other. Furthermore, we provide the details of the ROS nodes developed for controlling the robotic arm and gripper, given the information of a human’s bare hands sensed by the Leap Motion. Then, we demonstrate our system executing a simple pick-and-place task, and discuss possible benefits and costs of this HRI concept.

One of the most difficult challenges for terrestrial robotic platforms in disaster scenarios is their inability to traverse highly irregular terrain. Many different robotic architectures have been proposed over recent years, each with benefits and shortfalls. In this work, we propose a Polyurethane Foam depositing system, which can be applied to any such platform and increase its ability to overcome obstacles significantly. The system proposed is inexpensive, and the way in which it overcomes obstacles allows very simple control systems for autonomy. The deposited foam has a potential expansion ratio of over 33 $$\times $$ × its constituent parts and a final compressive strength exceeding 2 MPa, final mechanical properties can be tuned on board. The system has been implemented on a two-tracked rover and its autonomous responses tested against significant objects and chasms. The results show that the amount of foam deposited can be well controlled and multiple layers can be stacked on top of each other to significantly increase altitude.

This paper presents the design and development of Omni-Pi-tent, a self-reconfigurable modular robot capable of self-repair during continuous motion. This paper outlines the design features necessary for Dynamic Self-repair experiments and how the design of Omni-Pi-tent implements them, we summarise the construction of the first prototype and discuss initial experiments testing some of its key sensors and actuators. In addition, the paper describes experiments in which empirical data from laboratory tests of sensor hardware was integrated into V-REP simulations by means of creating custom sensor models so as to reduce the reality gap.

Robots are increasingly used in different scenarios, depending on the development of their capabilities and performance. The accelerating growth of robotics applications requires a tool that can comprehensively capture a wide range of robot capabilities. In this study, we evaluate robot capabilities using a structure known as “Towards Robot Characterization” (ToRCH) recently developed to meet this need. This structure defines robot capabilities and consequently enables capabilities and applications to be mapped against each other. An experiment was conducted to obtain the capabilities of two scenarios presented by the NAO robot. The method used to capture the capabilities was performed via the ToRCH structure. ToRCH implicitly illustrates the scenarios in a simple capability profile. This research assesses two aspects of the ToRCH capabilities capturing process. First, it verifies the moderate agreement level among roboticists in using ToRCH to capture the robot’s capabilities. Second, it demonstrates the richness of the ToRCH structure for capturing robot capabilities compared to the Multi-Annual Roadmap (MAR) levels. This initial study evaluates the ToRCH method in extracting different capability levels and illustrating them in a robot capability profile. It therefore highlights the potential of ToRCH in classifying robots.

In this paper the problem of the synthesis of optimal manoeuver trajectories for autonomous space vehicles and robots is revisited. It is shown that it is entirely feasible to construct optimal manoeuver trajectories from considerations of only the rigid body kinematics rather than the complete dynamics of the space vehicle or robot under consideration. Such an approach lends itself to several simplifications which allow the optimal angular velocity and translational velocity profiles to be constructed, purely from considerations of the body kinematic relations. In this paper the body kinematics is formulated, in general, in terms of the quaternion representation attitude and the angular velocities are considered to be the steering inputs. The optimal inputs for a typical attitude manoeuver is synthesized by solving for the states and co-states defined by a two point boundary value problem. A typical example of a space vehicle pointing problem is considered and the optimal torque inputs for the synthesis of a reference attitude trajectory and the reference trajectories are obtained.

This work demonstrates how dynamic robot behaviour that responds to different types of network disturbances can improve communication and mission performance in a Multi-Robot Team (MRT). A series of experiments are conducted which show how two different network perturbations (i.e. packet loss and signal loss) and two different network types (i.e. wireless local area network and ad-hoc network) impact communication. Performance is compared using two MRT behaviours: a baseline versus a novel dynamic behaviour that adapts to fluctuations in communication quality. Experiments are carried out on a known map with tasks assigned to a robot team at the start of a mission. During each experiment, a number of performance metrics are recorded. A novel dynamic Leader-Follower (LF) behaviour enables continuous communication through two key functions: the first reacts to the network type by using signal strength to determine if the robot team must commit to grouping together to maintain communication; and the second employs a special task status messaging function that guarantees a message is communicated successfully to the team members. The results presented in this work are significant for real-world multi-robot system applications that require continuous communication amongst team members.

This paper presents a novel autonomous air-hockey playing collaborative robot (cobot) that provides human-like gameplay against human opponents. Vision-based Bayesian tracking of the puck and striker are used in an Analytic Hierarchy Process (AHP)-based probabilistic tactical layer for high-speed perception. The tactical layer provides commands for an active control layer that controls the Cartesian position and yaw angle of a custom end effector. The active layer uses optimal control of the cobot’s posture inside the task nullspace. The kinematic redundancy is resolved using a weighted Moore-Penrose pseudo-inversion technique. Experiments with human players show high-speed human-like gameplay with potential applications in the growing field of entertainment robotics.

Search and Rescue robotics is a relatively new field of research, which is growing rapidly as new technologies emerge. However, the robots that are usually applied to the field are generally small and have limited functionality, and almost all of them rely on direct control from a local operator. In this paper, a novel wheeled Search and Rescue robot is proposed which considers new methods of controlling the robot, including using a wireless “tether” in place of a conventional physical one. A prototype is then built which acts as a proof of concept of the robot design and wireless control. The prototype robot is then evaluated to prove its mobility, wireless control and multi-hop networking. The experimental results demonstrate the effectiveness of the proposed design incorporating the rocker-bogie suspension system and the multi-hop method of “wireless tethering”.

Recent progress in soft robotics has seen new types of actuation mechanisms based on apical extension which allows robots to grow to unprecedented lengths. Eversion robots are a type of robots based on the principle of apical extension offering excellent maneuverability and ease of control allowing users to conduct tasks from a distance. Mechanical modelling of these robotic structures is very important for understanding their operational capabilities. In this paper, we model the eversion robot as a thin-walled cylindrical beam inflated with air pressure, using Timoshenko beam theory considering rotational and shear effects. We examine the various failure modes of the eversion robots such as yielding, buckling instability and lateral collapse, and study the payloads and operational limits of these robots in axial and lateral loading conditions. Surface maps showing the operational boundaries for different combinations of the geometrical parameters are presented. This work provides insights into the design of eversion robots and can pave the way towards eversion robots with high payload capabilities that can act from long distances.

The idea of this research is to evolve the shape of robots within a swarm, in order for them to work better as a whole. Small robots are not so powerful individually, but when cooperating with each other, by physically hooking together forming a larger organism for example, they may be able to solve more complex tasks. The shape each robot has influences the way they physically interact and, taking advantage of the morphological computation phenomenon, I show that evolving the robots’ morphology in a swarm makes it more efficient for the task of transporting objects, even in comparison to evolving the robot’s controller. In order to fulfill this objective, I have evolved the shape of arm-like structures for the robots’ bodies and their controller separately, and compared the results with control experiments.

Traditionally, wheeled rovers are used for planetary surface exploration and six-wheeled chassis designs based on the Rocker-Bogie suspension system have been tested successfully on Mars. However, it is difficult to explore craters and crevasses using large six or four-wheeled rovers. Innovative designs based on smaller Di-Wheel Rovers might be better suited for such challenging terrains. A Di-Wheel Rover is a self - balancing two-wheeled mobile robot that can move in all directions within a two-dimensional plane, as well as stand upright by balancing on two wheels.This paper presents the outcomes of a feasibility study on a Di-Wheel Rover for planetary exploration missions. This includes developing its chassis design based on the hardware and software requirements, prototyping, and subsequent testing. The main contribution of this paper is the design of a self-balancing control system for the Di-Wheel Rover. This challenging design exercise was successfully completed through extensive experimentation thereby validating the performance of the Di-Wheel Rover. The details on the structural design, tuning controller gains based on an inverted pendulum model, and testing on different ground surfaces are described in this paper. The results presented in this paper give a new insight into designing low-cost Di-Wheel Rovers and clearly, there is a potential to use Di-Wheel Rovers for future planetary exploration.

Situational awareness in remote environments is crucial for human operators to teleoperate mobile manipulators confidently and reliably. Visual feedback is the most common way for environment perception, providing rich information to human operators. This paper proposes an intuitive teleoperation method by combining gaze tracking and hand motion detection to teleoperate a mobile manipulator. A camera is fixed on the end-effector of the mobile robot’s arm to provide visual feedback, acting as the eye of the teleoperator. Rather than direct remote control of the robot, an adaptive auto-correction mechanism is introduced for helping human operators to achieve better hand-eye coordination of the teleoperation experiences. The mobile manipulator can adjust its behaviours, such as speed, while gaze and hand movements of the operator are in different states. The experiments carried out demonstrated the effectiveness of the proposed algorithm and the survey evaluation verified the practical application value of the system.

The problem of Unexploded Ordnance/Bomb (UXO/UXB) affects most big cities in UK and Europe. It is estimated that 10% of aerially dropped bombs failed to explode. The heavy weight of the bombs allowed them to penetrate the ground to a depth that may exceed 14 m. The disturbance of these bombs could result in fatal explosion and represents a serious danger to construction and foundations workers. Current methods for deep bomb detection include predrilled and pushed methods. The pushed method is the preferred technique but it cannot penetrate hard ground. The alternative is the predrilled method that employs a more powerful rotary drilling method to drill and scan the borehole in stages. However, it is very time-consuming and costly. This paper describes the development of an instrument equipped into a rotary drilling rig, which captures, transmits and records earth magnetic field wirelessly while drilling. The proposed system allows UXO detection in real time whilst drilling in various types of ground conditions and provides faster, safer and cheaper method of UXO detection overcoming the limitations of the existing methods. In order to further improve the safety of deeply buried UXB clearance in hazardous environment, this solution can be attached to an autonomous vehicle or robot while data can be remotely collected. Hence, this work focuses on designing a sensor system that can identify deeply buried underground bombs and can be integrated into a mobile robotic system. Multiple challenges are associated with the development of the proposed solution, mainly: strong magnetic noise interference caused by surrounding metal on the measurements, real time data transmission and the very harsh and noisy drilling environment. This paper addresses the telemetry challenge.

This work addresses the performance of several local planners for navigation of autonomous pallet trucks in the presence of humans in a simulated warehouse as well as a complementary approach developed within the ILIAD project. Our focus is to stress the open problem of a safe manoeuvrability of pallet trucks in the presence of moving humans. We propose a variation of ROS navigation stack that includes in the planning process a model of the human robot interaction.

Collaborative robots (Cobots) are indispensable tools in the factories of the future. Owing to their safety centered design, Cobots are allowed to work side by side with humans, making their use as an assistive third hand appealing for tedious assembly tasks. Consequently, we propose a robot that can be hand-guided to lift and hold parts in place while the human performs assembly tasks. Such functionality reduces the risk to workers (falling components for example), provides precision, allows lifting heavier parts, and increases productivity by allowing human workers to focus on more value-added tasks.

Swarm robotics is an emerging technology that has the potential to revolutionise precision agriculture by coordinating fleets of small autonomous vehicles to minimise soil damage, increase farming resolution, lower the cost of automation, and provide solutions that are intrinsically safer and more sustainable than large monolithic systems. Here, we propose a novel swarm robotic system for autonomous harvesting of cereal crops such as wheat and barley. In contrast to existing agricultural swarm robotic systems, we intend to use small autonomous versions of traditional agricultural vehicles, in an attempt to narrow the skills gap for future end-users.

This paper presents a distributed boundary detection method for a swarm of robots. It employs the cyclic-shape algorithm together with a local coordinate identification, thereby allowing boundary detection without bearing measurements. Each robot can communicate with, and estimate the distance to, its neighbouring robots. The method is validated using swarms of up to 45 Kilobot robots.

Much of robotics research aims to develop control solutions that exploit the machine’s dynamics in order to achieve an extraordinarily agile behaviour [1]. This, however, is limited by the use of traditional model-based control techniques such as model predictive control and quadratic programming. These solutions are often based on simplified mechanical models which result in mechanically constrained and inefficient behaviour, thereby limiting the agility of the robotic system in development [2]. Treating the control of robotic systems as a reinforcement learning (RL) problem enables the use of model-free algorithms that attempt to learn a policy which maximizes the expected future (discounted) reward without inferring the effects of an executed action on the environment.

This paper presents a routing strategy for UAVs that can be applied in conjunction with lower level collision avoidance methods. The strategy allows individual UAVs to route themselves in 2D space in order to avoid areas of high-density traffic. The proposed approach is explored in simulation. The results demonstrate a safer system operation when the routing strategy is used, compared with just a simple collision avoidance method.

Exploration is one of the fundamental problems in mobile robotics. Efforts to address this problem made over the past two decades divide into two approaches: reactive approaches, that make only instantaneous decisions, and map-based approaches involving e.g. grid, metric, or topological representations. Comparative studies have so far largely focused on comparing different map-based algorithms, while no common framework to compare them to purely reactive approaches currently exists. This paper aims at creating a framework to simulate, evaluate, and compare exploratory algorithms as different as reactive and map-based approaches. Preliminary results are demonstrated for two reactive algorithms, random walk and wall follower, and one map based approach, pheromone potential field, have been implemented. Measurements of navigation success, time to success, as well as computational and memory usage reveal a dominance of simple wall-following over the map-based potential field approach, and a distinct load/efficacy trade off for random walks. These preliminary results challenge the common assumptions that maps are needed for successful and efficient exploration and navigation.

Robotic technologies have been long-serving in space, and are still an active and ever-growing field of research. Satellite mounted manipulators allow more ambitious tasks to be carried out in a safer and more timely manner, by limiting the need for astronaut intervention during task execution. Downsizing these free-flying space robots will expand their potential by increasing their versatility, allowing task sharing between multiple systems, as well as further lowering mission costs and timescales. Limited research has been done in assessing the practical challenges involved in downsizing a space robot and its consequences on overall performance. This paper presents a system level analysis into deciding the optimum dimensions for a manipulator mounted on a small free-flying spacecraft. Simulation results show the effect of downsizing on the efficiency of the manipulator and the overall system.

This short paper describes an approach for collecting a dataset of hand’s pictures and training a Deep Learning network that could enable the iCub robot to count on its fingers using solely its own cameras. Such a skill, mimicking children’s habits, can support arithmetic learning in a baby robot, an important step in creating artificial intelligence for robots that could learn like children in the context of cognitive developmental robotics. Preliminary results show the approach is promising in terms of accuracy.

For general grasping, a strong lightweight and compact robot hand needs to be designed as a robot end effector. This paper describes the design of a 3D printed anthropomorphic robot hand that actuates flexible fingers using Twisted String Actuators (TSAs). A total of 6 of these actuators were fitted within a limited confined space, the palm of the hand. This gives the hand 6 Degrees of Freedom (DOFs) 2 in the thumb and 1 in each of the other fingers. A simple modular design was used which allows for rapid prototyping of different finger designs with respect to different requirements at low cost. In this paper, only power grasping is considered for simplicity. The hand is capable of performing both spherical and cylindrical grasps whereby the flexible nature of the fingers allows for forming around the geometry of a target object. The maximum holding load of the hand was found to be 10 kg in performance tests.

In the upcoming industrial revolution (Industry 4.0) automation and robotics play a central role. Humans and robots are expected to share the same workspace and work safely side by side. Consequently, various collaborative robots have been introduced to the market. Nevertheless, those robots are still limited in their reactions. In some cases they are restricted to reducing their working speed as a response to the proximity of humans or they initiate an emergency stop, particularly if a contact is detected. In this paper, our work on real-time human robot collision avoidance is presented. Unlike the existing solutions, in our method the robot is provided with agile reactivity to human presence. The system is engineered to achieve natural collision avoidance behavior. As a result, the robot acts with smooth avoidance motion upon the proximity of human, giving him/her the space required to do his/her work in shared tasks between a human co-worker and robot.

Power wheelchairs can increase independence by supporting the mobility of their users. However, severe disabilities of users can render controlling the wheelchair difficult, if not impossible, especially over longer periods of time. This paper describes a proposal for research into techniques that would improve the experience and quality of life of wheelchair users by reducing the cognitive burden introduced by repetitive and complicated navigation tasks and manoeuvres. This will be achieved by sharing the control between the user and an autonomous controller. A number of techniques will be used to achieve this aim. Simultaneous Localization and Mapping (SLAM) and topological mapping will be used for navigation between rooms while Computer Vision techniques will allow the (semi) automatic recognition of places in the user’s home, based on the detection and categorization of objects. Finally, medium to high level automation will be provided. This includes automatic and transparent assistance with tasks such as navigating through doorways but also autonomous navigation to specific locations using high level constructs (“take me to the kitchen table”).