Biomimetic and Biohybrid Systems

Despite many studies, the locomotion of bipedal legged robots is still not perfect. All the current robots are based on a humanoid model, which is not the unique one in Nature. In this paper we compare the natural bipedies in order to explore new tracks to improve robotic bipedal locomotion. This study starts with a short review of the historical bases of the biological bipedies to explain the differences between the structures of the human and bird bodies. The observations on the kinematics of bird walking describe a modular system that can be reproduced in robotics to take advantage of the bird leg versatility. For comparison purposes, a bird model is scaled up to have the same mass and the same height of the center of mass as a humanoid model. Simulation results show that the bird model can execute larger strides and stay on course, compared with the humanoid model. In addition the results confirm the functional decomposition of the bird system into the trunk and the thighs for the one part, and the distal part of the leg for the other part.

The ‘red-green’ pathway of the retina is classically recognized as one of the retinal mechanisms allowing humans to gather color information from light, by combining information from L-cones and M-cones in an opponent way. The precise retinal circuitry that allows the opponency process to occur is still uncertain, but it is known that signals from L-cones and M-cones, having a widely overlapping spectral response, contribute with opposite signs. In this paper, we simulate the red-green opponency process using a retina model based on linear-nonlinear analysis to characterize context adaptation and exploiting an image-processing approach to simulate the neural responses in order to track a moving target. Moreover, we integrate this model within a visual pursuit controller implemented as a spiking neural network to guide eye movements in a humanoid robot. Tests conducted in the Neurorobotics Platform confirm the effectiveness of the whole model. This work is the first step towards a bio-inspired smooth pursuit model embedding a retina model using spiking neural networks.

This paper presents an anthropomorphic design of a robotic finger with contact-aided cross four-bar (CFB) linkages. Anatomical study shows that finger joints have a complex structure formed by non-symmetric surfaces and usually produce complex movement than a simple revolute motion. The articular system of human hand is firstly investigated. Kinematics of a CFB mechanism is then analyzed and computer aided design of fixed and moving centrodes of CFB mechanism is presented. Gripping analysis of human hand shows two easily ignored components of a finger, fingernail and soft fingertip. Based on the range of motion of the joints of the most flexible thumb finger, a two-joint anthropomorphic finger is developed by using contact-aided CFB linkages which can also be used for joint design of prosthetic knee. Prototype of a two-fingered hand is manufactured by using 3D printing technology and gripping of a wide range of objects is tested.

This paper will outline why product designers are exploring making processes of the natural world and how this is of benefit to traditional product design practice. This type of experimental design work pushes the boundaries of conventional product design in which mass-manufacture efficiency drives the design and production process. Products, which use growth processes as fundamental to the making process, are increasingly becoming more feasible for end-user acquisition. This paper will provide two case study examples. These case studies contextualise these products and how they co-exist and contribute to the well-established design approaches of digital fabrication and co-creation.

This paper describes the integration of multiple sensory recognition models created by a Synthetic Autobiographical Memory into a structured system. This structured system provides high level control of the overall architecture and interfaces with an iCub simulator based in Unity which provides a virtual space for the display of recollected events.

Building a social robot that is able to interact naturally with people is a challenging task that becomes even more ambitious if the robots’ interlocutors are children involved in crowded scenarios like a classroom or a museum. In such scenarios, the main concern is enabling the robot to track the subjects’ social and affective state modulating its behaviour on the basis of the engagement and the emotional state of its interlocutors. To reach this goal, the robot needs to gather visual and auditory data, but also to acquire physiological signals, which are fundamental for understating the interlocutors’ psycho-physiological state. Following this purpose, several Human-Robot Interaction (HRI) frameworks have been proposed in the last years, although most of them have been based on the use of wearable sensors. However, wearable equipments are not the best technology for acquisition in crowded multi-party environments for obvious reasons (e.g., all the subjects should be prepared before the experiment by wearing the acquisition devices). Furthermore, wearable sensors, also if designed to be minimally intrusive, add an extra factor to the HRI scenarios, introducing a bias in the measurements due to psychological stress. In order to overcome this limitations, in this work, we present an unobtrusive method to acquire both visual and physiological signals from multiple subjects involved in HRI. The system is able to integrate acquired data and associate them with unique subjects’ IDs. The implemented system has been tested with the FACE humanoid in order to assess integrated devices and algorithms technical features. Preliminary tests demonstrated that the developed system can be used for extending the FACE perception capabilities giving it a sort of sixth sense that will improve the robot empathic and behavioural capabilities.

Densely packed silver nanoparticles (AgNPs) were produced as continuous films on non-conductive substrates in a site-selective manner. The formation of the AgNPs were directed by blue light using extract from spinach leaves acting as photo-reducing agent. This bio-inspired production of reduced ions nanofilms benefit applications where seed conductive layers are required for the manufacture of metal parts embedded into plastics.

Quadrupeds exhibit adaptive limb coordination to achieve versatile and efficient locomotion. In particular, the leg-trajectory changes in response to locomotion speed. The goal of this study is to reproduce this modulation of leg-trajectory and to understand the control mechanism underlying quadruped locomotion. We focus primarily on the modulation of stiffness of the leg because the trajectory is a result of the interaction between the leg and the environment during locomotion. In this study, we present a “TEGOTAE”-based control scheme to modulate the leg stiffness. TEGOTAE is a Japanese concept describing the extent to which a perceived reaction matches the expected reaction. By using the presented scheme, foot-trajectories were modified and the locomotion speed increased correspondingly.

To assess the responses of an identified optic-flow processing interneuron in the fly motion-vision pathway, the H1-cell, we performed semi-closed-loop experiments using a bio-hybrid two-wheeled robotic platform. We implemented a feedback-control architecture that established ‘wall following’ behaviour of the robot based on the H1-cell’s spike rate. The analysis of neuronal data suggests the spiking activity of the cell depends on both the momentary turning radius of the robot as well as the distance of the fly’s eyes from the walls of the experimental arena. A phenomenological model that takes into account the robot’s turning radius predicts spike rates that are in agreement with our experimental data. Consequently, measuring the turning radius using on-board sensors will enable us to extract distance information from H1-cell signals to further improve collision avoidance performance of our fly-robotic interface.

In earthworms, traveling waves of body contraction and elongation result in soft body locomotion. This simple strategy is called peristaltic locomotion. To mimic this kind of locomotion, we developed a compliant modular worm-like robot. This robot uses a cable actuation system where the actuating cable acts like the circumferential muscle. When actuated, this circumferential cable contracts the segment diameter causing a similar effect to the contraction due to the circumferential muscles in earthworms. When the cable length is increased, the segment diameter increases due to restoring forces from structural compliance. When the robot comes in contact with an external constraint (e.g., inner walls of a pipe) continued cable extension results in both slack in the cable and inefficiency of locomotion. In this paper we discuss a probabilistic approach to detect slack in a cable. Using sample distributions over multiple trials and naïve Bayes classifier, we can detect anomalies in sampled data which indicate the presence of slack in the cable. Our training set included data samples from pipes of different diameters and flat surfaces. This algorithm detected slack within ±0.15 ms of slack being introduced in the cable with a success rate of 75 %. We further our research in understanding reasons for failure of the algorithm and working towards improvements on our robot.

This paper investigates and compares the effectiveness of haptic and audio modality for navigation in low visibility environment using a sensory augmentation device. A second generation head-mounted vibrotactile interface as a sensory augmentation prototype was developed to help users to navigate in such environments. In our experiment, a subject navigates along a wall relying on the haptic or audio feedbacks as navigation commands. Haptic/audio feedback is presented to the subjects according to the information measured from the walls to a set of 12 ultrasound sensors placed around a helmet and a classification algorithm by using multilayer perceptron neural network. Results showed the haptic modality leads to significantly lower route deviation in navigation compared to auditory feedback. Furthermore, the NASA TLX questionnaire showed that subjects reported lower cognitive workload with haptic modality although both modalities were able to navigate the users along the wall.

In this article, we present our initial work on sequence prediction of a visual target by implementing a cortically inspired method, namely Hierarchical Temporal Memory (HTM). As a preliminary test, we employ HTM on periodic functions to quantify prediction performance with respect to prediction steps. We then perform simulation experiments on the iCub humanoid robot simulated in the Neurorobotics Platform. We use the robot as embodied agent which enables HTM to receive sequences of visual target position from its camera in order to predict target positions in different trajectories such as horizontal, vertical and sinusoidal. The obtained results indicate that HTM based method can be customized for robotics applications that require adaptation of spatiotemporal changes in the environment and acting accordingly.

The interdisciplinary character of Bio-Inspired Design (BID) has resulted in a plethora of approaches and methods that propose different types of design processes. Although sustainable, creative and complex system design processes are not mutually incompatible they do focus on different aspects of design. This research defines areas of focus for the development of computational tools to support biomimetics, technical problem solving through abstraction, transfer and application of knowledge from biological models. An overview of analysed literature is provided as well as a qualitative analysis of the main themes found in BID literature. The result is a set of recommendations for further research on Computer-Aided Biomimetics (CAB).

A neuromechanical simulation of a planar, bipedal walking robot has been developed. It is constructed as a simplified musculoskeletal system to mimic the biomechanics of the human lower body. The controller consists of a dynamic neural network with central pattern generators (CPGs) entrained by force and movement sensory feedback to generate appropriate muscle forces for walking. The CPG model is a two-level architecture, which consists of separate rhythm generator (RG) and pattern formation (PF) networks. The presented planar biped model walks stably in the sagittal plane without inertial sensors or a centralized posture controller or a “baby walker” to help overcome gravity. Its gait is similar to humans’ with a walking speed of 1.2 m/s. The model walks over small obstacles (5 % of the leg length) and up and down 5° slopes without any additional higher level control actions.

Hardware capable of 3D sensing, such as the Microsoft Kinect, has opened up new possibilities for low-cost computer vision applications. In this paper, we take the first steps towards unsupervised canine posture classification by presenting an algorithm to perform canine-background segmentation, using depth shadows and infrared data for increased accuracy. We report on two experiments to show that the algorithm can operate at various distances and heights, and examine how that effects its accuracy. We also perform a third experiment to show that the output of the algorithm can be used for k-means clustering, resulting in accurate clusters 83 % of the time without any preprocessing and when the segmentation algorithm is at least 90 % accurate.

While navigating their environments it is essential for autonomous mobile robots to actively avoid collisions with obstacles. Flying insects perform this behavioural task with ease relying mainly on information the visual system provides. Here we implement a bio-inspired collision avoidance algorithm based on the extraction of nearness information from visual motion on the hexapod walking robot platform HECTOR. The algorithm allows HECTOR to navigate cluttered environments while actively avoiding obstacles.

We describe the design of a novel commercial biomimetic brain-based robot, MIRO, developed as a prototype robot companion. The MIRO robot is animal-like in several aspects of its appearance, however, it is also biomimetic in a more significant way, in that its control architecture mimics some of the key principles underlying the design of the mammalian brain as revealed by neuroscience. Specifically, MIRO builds on decades of previous work in developing robots with brain-based control systems using a layered control architecture alongside centralized mechanisms for integration and action selection. MIRO’s control system operates across three core processors, P1-P3, that mimic aspects of spinal cord, brainstem, and forebrain functionality respectively. Whilst designed as a versatile prototype for next generation companion robots, MIRO also provides developers and researchers with a new platform for investigating the potential advantages of brain-based control.

The Hybrid Neuroprosthesis (HNP) is a hydraulically actuated exoskeleton and implanted Functional Electrical Stimulation (FES) system that has been designed and fabricated to restore gait to people with spinal cord injuries. The exoskeleton itself does not supply any active power, instead relying on an implanted FES system for all active motor torques. The exoskeleton instead provides support during quiet standing and stance phases of gait as well as sensory feedback to the stimulation system. Three individuals with implanted functional electrical stimulation systems have used the system to successfully walk short distances, but were limited in the flexion torques the stimulation system could provide.

In this study, deformation patterns of an adult male lower face are measured and analyzed for efficient face design for android robots. We measured flow vectors for 96 points on the right half of the lower face for 16 deformation patterns, which are selected from Ekman’s action units. Namely, we measured 16 flow vector fields of facial skin flow. The flow vectors were created by placing ink markers on the front of the face and then video filming various facial motions. A superimposed image of vector fields shows that each point moves in various directions. Principle component analysis was conducted on the superimposed vectors and the contribution ratio of the first principal component was found to be 86 %. This result suggests that each facial point moves almost only in one direction and different deformation patterns are created by different combinations of moving lengths. Based on this observation, replicating various kinds of facial expressions on a robot face might be easy because an actuation mechanism that moves a single facial surface point in one direction can be simple and compact.

In nature, Anticipatory Postural Adjustments (APAs) are actions that precede predictable disturbances with the goal of maintaining a stable body posture. Neither the structure of the computations that enable APAs are known nor adaptive APAs have been exploited in robot control. Here we propose a computational architecture for the acquisition of adaptive APAs based on current theories about the involvement of the cerebellum in predictive motor control. The architecture is applied to a simulated self-balancing robot (SBR) mounting a moveable arm, whose actuation induces a perturbation of the robot balance that can be counteracted by an APA. The architecture comprises both reactive (feedback) and anticipatory-adaptive (feed-forward) layers. The reactive layer consists of a cascade-PID controller and the adaptive one includes cerebellar-based modules that supply the feedback layer with predictive signals. We show that such architecture succeeds in acquiring functional APAs, thus demonstrating in a simulated robot an adaptive control strategy for the cancellation of a self-induced disturbance grounded in animal motor control. These results also provide a hypothesis for the implementation of APAs in nature that could inform further experimental research.

In this paper, we describe an optimization approach to the legged locomotion problem. We designed a software environment to manipulate parametrized robot models. This environment is a platform developed for future experiments and for educational robotics purpose. It allows to generate dynamic models and simulate them using a physics engine. Experiments can then be made with both morphological and controller optimization. Here we describe the environment, propose a simple open loop generic controller for legged robots and discuss experiments that were made on a robot corpus using a black-box optimization.

As opposed to insects, todays walking robots are typically not constructed to withstand crashes. Whereas insects use a multitude of sensor information and have self-healing abilities in addition, robots usually rely on few specialized sensors that are essential for operation. If one of the sensors fails due to a crash, the robot is unusable. Therefore, most technical systems require static stability at all times to avoid damages and to guarantee utilizability, whereas insects can afford occasional failures. Despite the failure tolerance, insects also possess adhesive, “sticky” pads and claws at their feet that allow them to cling to the substrate, thus reducing the need for static stability. Nevertheless, insects, in particular stick insects, have been studied intensively to understand the underlying mechanisms of their leg coordination in order to adapt it for the control of robots. This work exemplarily evaluates the static stability of a single stick insect during walking and the stability of a technical system that is controlled by stick insect - inspired coordination rules.

Rodents are able to navigate within dynamic environments by constantly adapting to their surroundings. Hippocampal place-cells encode the animals current location and fire in sequences during path planning events. Place-cells receive excitatory inputs from grid-cells whose metric system constitute a powerful mechanism for vector based navigation for both known and unexplored locations. However, neither the purpose or the behavioral consequences of such mechanism are fully understood. During early exploration of a maze with multiple discrimination points, rodents typically manifest a conflict-like behavior consisting of alternating head movements from one arm of the maze to the other be- fore making a choice, a behavior which is called vicarious trial and error (VTE). Here, we suggest that VTE is modulated by the learning process between spatial- and reward-tuned neuronal populations. We present a hippocampal model of place- and grid-cells for both space representation and mental travel that we used to control a robot solving a foraging task. We show that place-cells are able to represent the agents current location, whereas grid-cells encode the robots movement in space and project their activity over unexplored paths. Our results suggest a tight interaction between spatial and reward related neuronal activity in defining VTE behavior.

This paper discusses the implementation of insect-inspired visual navigation strategies in flying robots, in particular focusing on the impact of changing height. We start by assessing the information available at different heights for visual homing in natural environments, comparing results from an open environment against one where trees and bushes are closer to the camera. We then test a route following algorithm using a gantry robot and show that a robot would be able to successfully navigate a route at a variety of heights using images saved at a different height.

USER (Universal SEmantic Representation) is a bio-inspired module implemented in a system on a chip (SoC), which builds a link between multichannel perception and semantic representation. The input data are projected into a generic bioinspired higher dimensional non-linear semantic space with high sparsity. A pooling of these semantic representations (global, dynamic and structural) is done automatically by a set of dynamic attractors embedding spatio-temporal histograms, being drastically more efficient than back-propagation. A supervised learning is used to build the association between the invariant multimodal semantic representations (histogram results) and the labels (‘words’). The invariant recognition is achieve thanks to multichannel multiscale dynamic attractors and bilinear representations - imitating brain attentional processes. USER modules can be cascaded, allowing to work at different levels of abstraction (or complexity). Due to its low consumption, small size and minimal price, USER targets deep learning, robotics, and Internet of Things (IoT) applications.

We present a multi-modal robot that flies, perches and climbs on outdoor surfaces such as concrete or stucco walls. Although the combination of flying and climbing mechanisms in a single platform extracts a weight penalty, it also provides synergies. In particular, a small amount of aerodynamic thrust can substantially improve the reliability of perching and climbing, allowing the platform to maneuver on otherwise risky surfaces. The approach is inspired by thrust-assisted perching and climbing observed in various animals including flightless birds.

This paper presents the EU EASEL project, which explores the potential impact and relevance of a robot in educational settings. We present the project objectives and the theorectical background on which the project builds, briefly introduce the EASEL technological developments, and end with a summary of what we have learned from the evaluation studies carried out in the project so far.

This paper outlines the development of a four-part needle prototype inspired by the ovipositor of parasitic wasps. In the wasp ovipositor, three longitudinal segments called valves move reciprocally to gain depth in the substrate. It has been suggested that serrations located along the wasp ovipositor induce a friction difference between moving and anchoring valves that is needed for this reciprocal motion. Such an anchoring mechanism may not be desired in a medical setting, as serrations can induce tissue damage. Our aim was to investigate whether a multipart needle can penetrate tissue phantom material with near-zero net push force while using needle parts devoid of surface gripping textures or serrations. Accordingly, a four-part needle prototype was developed and tested in gelatine substrates. The performance of the prototype was assessed in terms of the degree of slipping of the needle with respect to the gelatine, with less slip implying better performance. Slip decreased with decreasing gelatine concentration and increasing offset between the needle parts. Motion through gelatine was achieved with a maximum push force of 0.035 N. This study indicates the possibility of needle propagation into a substrate with low net push force and without the need of serrations on the needle surface.

Imaging systems are widely used in robotic systems to detect features of their surroundings and their position in order to guide the robot’s motion. In the paper, a variant of an artificial imaging system based on the unique bifocal eye of sunburst diving beetle (Thermonectus marmoratus) larvae is proposed. The biologically inspired imaging system of a single sensor and a coaxial lens form a superposition of two focused narrow and wide view angle images. The output image contains a high resolution area of interest and its periphery. The scanning motion of the bifocal imaging system is also imitated and provides positional relations between objects. Acquired images are used for distance assessment. The intended use of the proposed imaging system is in a guidance system of an autonomously moving robot with biologically inspired locomotion.

Praying mantises are excellent models for studying directed motion. They may track prey with rapid saccades of the head, prothorax, and legs, or actively pursue prey, using visual input to modulate their walking patterns. Here we present a conductance-based neural controller for MantisBot, a 28 degree-of-freedom robot, which enables it to use faux-visual information from a head sensor to either track or pursue prey with its prothorax and appropriate movements of one of its legs. The controller can switch between saccades and smooth tracking, as seen in pursuit, modulating only two neurons in its model thoracic ganglia via descending commands. Similarly, the neural leg controller redirects the direction of locomotion, and automatically produce reflex reversals seen in other insects when they change direction, via two simple descending commands.

Two main classes of reflexes relying on the vestibular system are involved in the stabilization of the human gaze: the vestibulocollic reflex (VCR), which stabilizes the head in space and the vestibulo-ocular reflex (VOR), which stabilizes the visual axis to minimize retinal image motion. Together they keep the image stationary on the retina.In this work we present the first complete model of eye-head stabilization based on the coordination of VCR and VOR. The model is provided with learning and adaptation capabilities based on internal models. Tests on a simulated humanoid platform replicating torso disturbance acquired on human subject performing various locomotion tasks confirm the effectiveness of our approach.

Robots are gradually but steadily being introduced in our daily lives. A paramount application is that of education, where robots can assume the role of a tutor, a peer or simply a tool to help learners in a specific knowledge domain. Such endeavor posits specific challenges: affective social behavior, proper modelling of the learner’s progress, discrimination of the learner’s utterances, expressions and mental states, which, in turn, require an integrated architecture combining perception, cognition and action. In this paper we present an attempt to improve the current state of robots in the educational domain by introducing the EASEL EU project. Specifically, we introduce the EASEL’s unified robot architecture, an innovative Synthetic Tutor Assistant (STA) whose goal is to interactively guide learners in a science-based learning paradigm, allowing us to achieve such rich multimodal interactions.

Aplysia californica is presented as a novel source of actuator and scaffold material for biohybrid robots. Collagen isolated from the Aplysia skin has been fabricated into gels and electrocompacted scaffolds. Additionally, the I2 muscle from the Aplysia buccal mass had been isolated for use as an organic actuator. This muscle has been characterized and the maximum force was found to be 58.5 mN with a maximum muscle strain of 12 ± 3 %. Finally, a flexible 3D printed biohybrid robot has been fabricated which is powered by the I2 muscle and is capable of locomotion at 0.43 cm/min under field stimulation.

Drosophila melanogaster has been studied to gain insight into relationships between neural circuits and learning behaviour. To test models of their neural circuits, a robot that mimics D. melanogaster larvae has been designed. The robot is made from silicone by casting in 3D printed moulds with a pattern simplified from the larval muscle system. The pattern forms air chambers that function as pneumatic muscles to actuate the robot. A pneumatic control system has been designed to enable control of the multiple degrees of freedom. With the flexible body and multiple degrees of freedom, the robot has the potential to resemble motions of D. melanogaster larvae, although it remains difficult to obtain accurate control of deformation.

Reviewing recent closely related developments at the crossroads of biomedical engineering, artificial intelligence and biomimetic technology, in this paper, we attempt to distinguish phenomenological consciousness into three categories based on embodiment: one that is embodied by biological agents, another by artificial agents and a third that results from collective phenomena in complex dynamical systems. Though this distinction by itself is not new, such a classification is useful for understanding differences in design principles and technology necessary to engineer conscious machines. It also allows one to zero-in on minimal features of phenomenological consciousness in one domain and map on to their counterparts in another. For instance, awareness and metabolic arousal are used as clinical measures to assess levels of consciousness in patients in coma or in a vegetative state. We discuss analogous abstractions of these measures relevant to artificial systems and their manifestations. This is particularly relevant in the light of recent developments in deep learning and artificial life.

The aim of this paper is to investigate a new approach to decode sensor information into spatial information. The tactile fingertip (TacTip) considered in this work is inspired from the operation of dermal papillae in the human fingertip. We propose an approach for interpreting tactile data consisting of a preprocessing dimensionality reduction step using principal component analysis and subsequently a regression model using a Gaussian process. Our results are compared with a classification method based on a biomimetic approach for Bayesian perception. The proposed method obtains comparable performance with the classification method whilst providing a framework that facilitates integration with control strategies, for example to perform controlled manipulation.

In order to survive in an unpredictable and changing environment, an agent has to continuously make sense and adapt to the incoming sensory information and extract those that are behaviorally relevant. At the same time, it has to be able to learn to associate specific actions to these different percepts through reinforcement. Using the biologically grounded Distributed Adaptive Control (DAC) robot-based neuronal model, we have previously shown how these two learning mechanisms (perceptual and behavioral) should not be considered separately but are tightly coupled and interact synergistically via the environment. Through the use of a simulated setup and the unified framework of the DAC architecture, which offers a pedagogical model of the phases that form a learning process, we aim to analyze this perceptual-behavioral binomial and its effects on learning.

Human robot interaction (HRI) often considers the human impact of a robot serving to assist a human in achieving their goal or a shared task. There are many circumstances though during HRI in which a robot may make errors that are inconvenient or even detrimental to human partners. Using the ROBOtic GUidance and Interaction DEvelopment (ROBO-GUIDE) model on the Pioneer LX platform as a case study, and insights from social psychology, we examine key factors for a robot that has made such a mistake, ensuring preservation of individuals’ perceived competence of the robot, and individuals’ trust towards the robot. We outline an experimental approach to test these proposals.

The Expressive Agents for Symbiotic Education and Learning project explores human-robot symbiotic interaction with the aim to understand the development of symbiosis over long-term tutoring interactions. The final EASEL system will be built upon the neurobiologically grounded architecture - Distributed Adaptive Control. In this paper, we present the design of an interaction scenario to support development of the DAC, in the context of a synthetic tutoring assistant. Our humanoid robot, capable of life-like simulated facial expressions, will interact with children in a public setting to teach them about exercise and energy. We discuss the range of measurements used to explore children’s responses during, and experiences of, interaction with a social, expressive robot.

The function of the human fingertip has been often debated. There have been studies focused on how the fingerprint affects the perception of high temporal frequencies, such as for improved texture perception. In contrast, here we focus on the effects of the fingerprint on the spatial aspects of tactile perception. We compare two fingertips, one with a biomimetic fingerprint and the other having a smooth surface. Tactile data was collected on a sharp edged stimulus over a range of locations and orientations, and also over a smooth (cylindrical) object. The perceptual capabilities of both (fingerprint and smooth) sensor types were compared with a probabilistic classification method. We find that the fingerprint increases the perceptual acuity of high spatial frequency features (edges) by 30–40% whilst not influencing the tactile acuity of low spatial frequency features (cylinder). Therefore the biomimetic fingerprint acts as an amplifier of high spatial frequencies, and provides us with evidence to suggest that the perception of high spatial frequencies is also one of the functions of the human fingertip.

Anticipating synchronisation (AS) is an interesting phenomenon whereby a dynamical system can trace the future evolution of an analogous ‘master’ system using delayed feedback. Although some have theorised that AS (or a similar phenomenon) has a role in human motor control, research has primarily focused on demonstrating it in novel systems rather than its practical applications. In this paper we explore one such application: by coupling the dynamics of the joints in a simulated robot arm, we seek to demonstrate that AS can have a functional role in motor control by reducing instability during a reaching task.

This paper presents the design of a vision system for MantisBot, a robot designed to model a male Tenodera sinensis. The vision system mounted on the head of the robot consists of five oriented voltaic panels that distinguish the location of the most dominant light source in reference to the head’s location. The purpose of this vision system is to provide the neural control system of the robot with descending commands, allowing for the investigation of transitions between targeted motions. In order to mimic the vision system of the insect, the voltaic panels were positioned in a particular orientation that when combined with a designed electric circuit yields an output that can be used by the neural controller to control the robot’s motion. This paper summarizes the design of the vision system and its outputs. It also presents calibration data revealing the system’s ability to encode a light source’s elevation and azimuth angle as well as its distance.

Advanced tactile capabilities could help new generations of robots work co-operatively with people in a wider sphere than these devices have hitherto experienced. Robots could perform autonomous manipulation tasks and exploration of their environment. These applications require a thorough characterisation of the force measurement capabilities of tactile sensors. For this reason, this work focuses on the characterisation of the force envelope of the biomimetic, low-cost and robust TacTip sensor. Comparison with a traditional load cell shows that when identifying low forces and changes in position the TacTip proves significantly less noisy.

Insects exhibit surprisingly adaptive and versatile locomotion despite their limited computational resources. Such locomotor patterns are generated via coordination between leg movements, i.e., an interlimb coordination mechanism. The clarification of this mechanism will lead us to elucidate the fundamental control principle of animal locomotion as well as to realize truly adaptive legged robots that could not be developed solely by conventional control theory. In this study, we tried to model the interlimb coordination mechanism underlying hexapod locomotion on the basis of a concept called “TEGOTAE,” a Japanese concept describing how well a perceived reaction matches an expectation. Preliminary experimental results show that our proposed TEGOTAE-based control scheme allows us to systematically design a decentralized interlimb coordination mechanism that can well-reproduce insects’ gait patterns.

Legged animals exhibit adaptive and resilient locomotion through their inter-limb coordination. Our long-term goal of this study is to develop a systematic design scheme for legged robots by elucidating the inter-limb coordination mechanism of various legged animals from a unified viewpoint. As a preliminary step towards this, we here focus on millipedes. We performed behavioral experiments on a terrain with gap, and found that legs do not tend to move without the ground contact. Based on this qualitative finding, we proposed a decentralized control scheme using local force feedback.

Snakes exhibit “scaffold-based locomotion” wherein they actively utilize terrain irregularities and move effectively by pushing their body against the scaffolds that they encounter. Implementing the underlying mechanism in snake-like robots will enable them to work well in unstructured real-world environments. In this study, we proposed a decentralized control scheme for snake-like robots based on TEGOTAE, a Japanese concept describing how well a perceived reaction matches an expectation, to reproduce scaffold-based locomotion. A preliminary experimental result showed that reaction forces from environment are evaluated based on TEGOTAE in real time and those beneficial for propulsion of the body are selectively exploited.

Being able to selectively attend to stimuli is essential for any agent operating in noisy environments. Humans have developed, or are inherently equipped, with mechanisms to do so; these mechanisms are a rich source of debate. Here, we contribute to it by building a functional model to describe the findings of an existing psychophysiological experiment demonstrating how early and late saccades as well as “conscious report” are affected by varying levels of cognitive load. The model adheres to the established principles of neurophysiology. In a task focused on the monitoring of moving stimuli, where objects usually move predictably but randomly deviate from it every 0.2 s, change in cognitive load is reflected in the proportion of late saccades and behavioural reports in response to the task. It also provides evidence that the physiological structure of the brain is capable of implementing selective attention using a method other than the attentional spotlight.

This paper focus on bio-robots driven by cardiomyocyte-powered actuators. Towards biomaterial-based robotic actuators that exhibit co-operative motion, this study intends to deal with mutual entrainment between two cardiac-oscillators with mechanical interaction. The cardiac-oscillator is an oriented collagen sheet seeded cultured rat cardiomyocytes. Mechanically coupled cardiac-oscillator sysytem was developed consisting of two cardiac-oscillators connected mechanically via a PDMS film. Mechanically coupled cardiac-oscillator system and a single cardiac-oscillator was applied electric stimulation for investigating the function of mechanical structure. As a result of experiment, we observed mutual entrainment occurred in only mechanically coupled structure.

Despite the appealing concept of “central pattern generator” (CPG)-based control for bipedal walking, there is currently no systematic methodology for designing a CPG controller. To tackle this problem, we employ a unique approach: We attempt to design local controllers in the CPG model for bipedal walking based on the viewpoint of “TEGOTAE”, which is a Japanese concept describing how well a perceived reaction matches an expectation. To this end, we introduce a TEGOTAE function that quantitatively measures TEGOTAE. Using this function, we can design decentralized controllers in a systematic manner. We designed a two-dimensional bipedal walking model using TEGOTAE functions and constructed simulations using the model to verify the validity of the proposed design scheme. We found that our model can stably walk on flat terrain.

We developed a tactile vision sensor, by converting light intensity from narrowband reflection spectroscopy into different stimuli such as sound. This creates the perception of colours though alternative senses. The device extends work done to develop a wearable sensor of erythema and shows the outcome from the collaboration between scientists and artist. It discovers new applications for existing research combining technology and design.

Humans use a contour following exploratory procedure to estimate object shape by touch. Here we demonstrate autonomous robotic contour following with a biomimetic tactile fingertip, the TacTip, using an active touch method previously developed for other types of touch sensors. We use Bayesian sequential analysis for perception and implement an active control strategy to follow an object contour. The technique is tested on a 110 mm diameter circle and yields results comparable with those previously achieved for other tactile sensors. We intend to extend the work onto a different robot platform with improved trajectory control to improve robustness, speed and match with human performance.

Robots, as well as machine learning algorithms, have proven to be, unlike human beings, very sensitive to errors and failure. Artificial intelligence and machine learning are nowadays the main source of algorithms that drive cognitive robotics research. The advances in the fields have been huge during the last year, beating expert-human performance in video games, an achievement that was unthinkable a few years ago. Still, performance has been assessed by external measures not necessarily fit to the problem to solve, what lead to shameful failure on some specific tasks. We propose that the way to achieve human-like robustness in performance is to consider the self of the agent as the real source of self-evaluated error. This offers a solution to acting when information or resources are scarce and learning speed is important. This paper details our extension of the cognitive architecture DAC to control embodied agents and robots, through self-generated signals, from needs, drives, self-generated value and goals.

Biomimetic tactile sensors often need a large amount of training to distinguish between a large number of different classes of stimuli. But when stimuli vary in one continuous property such as sharpness, it is possible to reduce training by using a discrimination approach rather than a classification approach. By presenting a biomimetic tactile sensing device, the TacTip, with a single exemplar of edge sharpness, the sensor was able to discriminate between unseen stimuli by comparing them to the trained exemplar. This technique reduces training time and may lead to more biologically relevant models of perceptual learning and discrimination.

In this paper, we discuss body-induced motion of quadraped walking, in the presence of rolling contact between its feet and the ground. This paper is based on the authors’ previous study, where we had shown that active “rocking” of the torso from the left to the right (in the lateral plane) would induce alternative swing of the legs (in the sagittal plane), which result in autonomous forward walking. Now in this paper, we focus on the influence of the rolling contact of the feet and the ground; we point out that the forwading displacement is the sum of “walking effect” and “rolling effect”, and the rolling effect can be observed even when each pair of the fore legs and the hind legs are bound together. Moreover, in order to extract pure rolling motion, we analyze in detail the model of two hemispheres connected via a spine rod, as an extremely reduced model of quadruped locomotion.

Fish have developed advanced hydrodynamic sensing capabilities using neuromasts, a series of collocated inertial sensors distributed over their body. We have developed the hydromast, an upscaled version of this sensing modality in order to facilitate near bed sensing for aquatic systems. Here we introduce the concept behind this bioinspired flow sensing device as well as the first results from laboratory investigations.

In this work in progress report we present Remora, a system for designing interactive subdermal devices. Remora builds upon methods and technologies developed by body modification artists. The development has so far focussed on battery and power management as well as redundancy features. Remora consists of a series of hardware modules and a corresponding software environment. Remora is designed for body modification artists to design their own implants, but will also be a platform for researchers interested in sub-dermal interaction and hybrid systems. We have so far implemented a prototype device; future work will include in depth evaluation of this device as well as user studies, a graphical development environment and additional hardware and software modules.

Not only 4-legged animals but also 6-legged creatures exhibit a range of gaits. To reveal the mechanism underlying the gait of animals and realize such a gait in a multi-legged robot, it is important to understand the inter-limb coordination. It is possible that the inter-limb coordination is strongly influenced by the structure of trunk to which each leg is connected. In this study, we designed a brand-new insect-inspired trunk mechanism that is equipped with an actuator. Using this trunk, we built a 6-legged robots with passive limbs. We performed walking experiments with the developed robot and analyzed its gait, especially the relationship between the structure of the trunk and the walking velocity. The experimental result shows an insect inspired body structure affects its gait and the walking velocity of 6-legged walking robot.

The manner in which quadrupeds change their locomotion patterns with change in speed is poorly understood. In this paper, we demonstrate spontaneous gait transition by using a quadruped robot model with a head segment, for which leg coordination can be self-organized through a simple “central pattern generator” (CPG) model with a postural reflex mechanism. Our model effectively makes use of head movement for the gait transition, suggesting that head movement is crucial for the reproduction of the gait transition to high-speed gaits in quadrupeds.

We introduce a method for generalizing tactile features across different orientations and locations, inspired by recent studies demonstrating tactile generalization in humans. This method is applied to two 3d-printed bioinspired optical tactile sensors. Internal pins acting as taxels are arranged with rotational and translational symmetry in these sensors. By rotating or translating a small sample of tactile images, we are able to generalize tactile stimuli to new orientations or locations along the sensor respectively. Applying these generalization methods in combination with active perception leads to the natural formation of a fovea of accurate real tactile data surrounded by moderately less accurate generalized data used to focus the sensor’s tactile perception.

Centipedes exhibit adaptive locomotion via coordination of their numerous legs. In this study, we aimed to clarify the inter-limb coordination mechanism by focusing on autonomous decentralized control. Based on our working hypothesis that physical interaction between legs via the body trunk plays an important role for the inter-limb coordination, we constructed a model wherein each leg is driven by a simple local reflexive mechanism.

Our goal is to develop snake-like robots that can exhibit versatile locomotion patterns in response to the environments like real snakes. Towards this goal, in our other work, we proposed an autonomous decentralized control scheme on the basis of a concept called TEGOTAE, a Japanese concept describing how well a perceived reaction matches an expectation, and then succeeded in reproducing scaffold-based locomotion, a locomotion pattern observed in unstructured environments, via real-world experiments. In this study, we demonstrated via simulations that concertina locomotion observed in narrow spaces can be also reproduced by using the proposed control scheme.