Biomimetic and Biohybrid Systems

Simple animals adapt with impunity to the most challenging of conditions without training or supervision. Their behavioral repertoire is organized into a layered set of exteroceptive reflexes that can operate in parallel and form sequences in response to affordances of the environment. We have developed a common architecture that captures these underlying mechanisms for implementation in engineered devices. The architecture instantiates the underlying networks with discrete time map-based neurons and synapses on a sequential processor. A common board set instantiates releasing mechanisms, command neurons, coordinating neurons, central pattern generators, and reflex functions that are programmed as networks rather than as algorithms. Layered exteroceptive reflexes mediate heading control, impediment compensation, obstacle negotiation, rheotaxis, docking, and odometry and can be adapted to a variety of robotic platforms. We present the implementation of this architecture for three locomotory modes: swimming, walking, and flying.

The cerebellum is a neuronal structure comprising half the neurons of the central nervous system. It is essential in motor learning and classical conditioning. Here we present a digital electronic module, pluggable to an artificial autonomous system, designed following the neural structure of the cerebellum. It emulates the associative learning function as described in the context of classical conditioning. Building on our previous work we propose a neuromorphic implementation portable to a Field Programmable Gate Array (FPGA), capable of generating responses of variable amplitude. To validate our design we test it with the simulation of a robot performing a navigation task on a curvy track. Our digital cerebellum is able to make adaptively-timed rotations with variable amplitude suitable for the track. This suggests that the Purkinje cell dependent learning circuits of the cerebellum do not only time the triggering of actions but can also tune the specific response amplitude.

Many of the processes involved into the synthesis of human motion have much in common with problems found in robotics research. This paper describes the modeling and the simulation of a novel bipedal robot based on Series Elastic Actuators (SEAs) [1]. The robot model takes inspiration from the human musculoskeletal organization. The geometrical organization of the robot artificial muscles is based on the organization of human muscles. In this paper we study how the robot active and passive elastic actuation structures develop force during selected motor tasks. We then compare the robot dynamics to that of the human during the same motor tasks. The motivation behind this study is to translate the mechanisms underlying the human musculoskeletal dynamics to the robot design stage for the purpose of developing machines with better motor abilities and energy saving performances.

In this paper, we investigate control architectures that combine implicit models of behavior with ubiquitous sensory input, for soft hyper-redundant robots. Using a Wilson-Cowan neuronal model in a continuum arrangement that mirrors the arrangement of muscles in an earthworm, we can create a wide range of steady waves with descending signals. Here, we demonstrate how sensory feedback from individual segment strains can be used to modulate the behavior in desirable ways.

In this article, we propose a solution to the underwater exploration of objects using a new sensor inspired from the electric fish. The solution is free of any model and is just based on the combination of elementary behaviors, each of these behaviors being achieved through direct feedback of the electric measurements. The solution is robust, cheap and easy to implement. After, stating and interpreting it, the article ends with a few experimental results consisting in exploring small and large unknown objects.

Rodents have been widely studied for their adaptive navigation capabilities. They are able to exhibit multiple navigation strategies; some based on simple sensory-motor associations, while others rely on the construction of cognitive maps. We previously proposed a computational model of parallel learning processes during navigation which could reproduce in simulation a wide set of rat behavioral data and which could adaptively control a robot in a changing environment. In this previous robotic implementation the visual approach (or taxon) strategy was however paying attention to the intra-maze landmark only and learned to approach it. Here we replaced this mechanism by a more realistic one where the robot autonomously learns to select relevant landmarks. We show experimentally that the new taxon strategy is efficient, and that it combines robustly with the planning strategy, so as to choose the most efficient strategy given the available sensory information.

Electrical control of optical focalisation is important in several fields, such as consumer electronics, medical diagnostics and optical communications. As an alternative to complex, bulky and expensive current solutions based on shifting constant-focus lenses, here we report on an electrically tunable lens made of dielectric elastomers as ‘artificial muscle’ materials. The device is inspired to the architecture of the crystalline lens and ciliary muscle of the human eye. A fluid-filled elastomeric lens is integrated with an annular elastomeric actuator that works as an artificial muscle. Electrical activation of the artificial muscle deforms the lens, with a relative variation of focal length comparable to that of the human lens. Optical performance is achieved with compact size, low weight, fast and silent operation, shock tolerance, no overheating, low power consumption, and inexpensive off-the-shelf materials. Results show that combing bio-inspired design with dielectric elastomer artificial muscles can open new perspectives on tunable optics.

Despite the increasing mutual interest, robotics and cognitive sciences are still lacking common research grounds and comparison methodologies, for a more efficient use of modern technologies in aid of neuroscience research. We employed our humanoid robot for reproducing experiments on saccadic adaptation, on the same experimental setup used for human studies. The behavior of the robot, endowed with advanced sensorimotor skills and high autonomy in its interaction with the surrounding environment, is based on a model of cortical sensorimotor functions. We show how the comparison of robot experimental results with human and computational modeling data allows researchers to validate and assess alternative models of psychophysical phenomena.

This paper describes the design and control of a suspension based on electroactive polymers for controlling the landing dynamics of a jumping robot. Tunable suspension elements can electrically change their stiffness up to a factor of 10 in less than 0.01 seconds. We discuss design parameters and performance relevant to bio-inspired systems and demonstrate the ability to operate in positive (actuator), neutral (spring-like), or negative (damping or braking) workloops. When applied to a single-legged robot, positive workloops allow sustained periodic hopping while negative workloops can be used to rapidly achieve equilibrium during a landing event, acting in a similar manner to muscle in jumping animals. Extended bio-inspired applications are discussed.

Biological principles of closed-loop motor control have gained much interest over the last years for their potential applications in robotic system. Although some progress has been made in understanding of how biological systems use sensory signals to control reflex and voluntary behaviour, experimental platforms are still missing which allow us to study sensorimotor integration under closed-loop conditions. We developed a fly-robot interface (FRI) to investigate the dynamics of a 1-DoF image stabilization task. Neural signals recorded from an identified visual interneuron were used to control a two-wheeled robot which compensated for wide-field visual image shifts caused by externally induced rotations. We compared the frequency responses of two different controllers with static and adaptive feedback gains and their performance and found that they offer competing benefits for visual stabilization. In future research will use the FRI to study how different sensor systems contribute towards robust closed-loop motor control.

Desert ants when foraging for food, navigate by performing path integration and exploiting landmarks. In an earlier paper, we proposed a decentralized neurocontroller that describes this navigation behavior. As by real ants, landmarks are recognized depending on the context, i.e. only when landmarks belong to the path towards the current goal (food source, home). In this earlier version, neither position nor quality of the food sources can be learnt, the memory is preset. In this article, we present a new version, whose memory elements allow for learning food place vectors and quality. When the agent meets a food source, it updates the quality value, if this source is already known, or stores position and quality, if the source is new. Quality values are used to select food sources to be visited. When one source has a too low quality, the agent also finds a shortcut to another known food source.

The honeybee dance communication system is one of the most intriguing examples of information transfer in the animal kingdom. After returning from a valuable food source honeybee foragers move vigorously, in a highly stereotypical pattern, on the comb surface conveying polar coordinates of the field site to a human observer. After 60 years of intense research it remains still unknown how the bees decode the dance. To resolve this question we have built a robotic honeybee that is able to reproduce all stimuli found to be generated in the dance ([12]). By imitating single stimuli or combinations and tracking the bees’ ensuing behavior we are able to identify essential signals in the communication process. In this paper we describe the design of our current prototype, show how we validated the function of the robotic wing buzzes and propose a reactive behavior control on the basis of relative body configurations of nearby bees measured by custom smart camera modules. We will conclude by showing first promising result of field experiments within a live honeybee colony.

Recently, we have proposed that the dense local and sparse long-range connectivity of the visual cortex accounts for the rapid and robust transformation of visual stimulus information into a temporal population code, or TPC. In this paper, we combine the canonical cortical computational principle of the TPC model with two other systems: an attention system and a hippocampus model. We evaluate whether the TPC encoding strategy can be efficiently used to generate a spatial representation of the environment. We benchmark our architecture using stimulus input from a real-world environment. We show that the mean correlation of the TPC representation in two different positions of the environment has a direct relationship with the distance between these locations. Furthermore, we show that this representation can lead to the formation of place cells. Our results suggest that TPC can be efficiently used in a high complexity task such as robot navigation.

Experimental research on decision making has been mainly focused on binary perceptual tasks. The generally accepted models describing the decision process in these tasks are the integrator models. These models suggest that perceptual evidence is accumulated over time until a decision is made. Therefore, the final decision is based solely on recent perceptual information. In behaviorally more relevant tasks such as foraging, it is however probable, that the current choice also depends on previous experience. To understand the implications of considering previous experience in an integrator model we investigate it using a cognitive architecture (DAC9) with a robot performing foraging tasks. Compared to an instantaneous decision making model we show that an integrator model improves performance and robustness to task complexity. Further we show that it compresses the information stored in memory. This result suggests a change in the way actions are retrieved from memory leading to self-generated actions.

Animal behaviour is rich, varied, and smoothly integrated. One plausible model of its generation is that behavioural sub-systems compete to command effectors. In small terrestrial mammals, many behaviours are underpinned by foveation, since important effectors (teeth, tongue) are co-located with foveal sensors (microvibrissae, lips, nose), suggesting a central role for foveal selection and foveation in generating behaviour. This, along with research on primate visual attention, inspires an alternative hypothesis, that integrated behaviour can be understood as sequences of foveations with selection being amongst foveation targets based on their salience. Here, we investigate control architectures for a biomimetic robot equipped with a rodent-like vibrissal tactile sensing system, explicitly comparing a salience map model for action guidance with an earlier model implementing behaviour selection. Both architectures generate life-like action sequences, but in the salience map version higher-level behaviours are an emergent consequence of following a shifting focus of attention.

From using forks to eat to maneuvering high-tech gadgets of modern times, humans are adept in swiftly learning to use a wide range of tools in their daily lives. The essence of ‘tool use’ lies in our gradual progression from learning to act ‘on’ objects to learning to act ‘with’ objects in ways to counteract limitations of ‘perceptions, actions and movements’ imposed by our bodies. At the same time, to learn both “cumulatively” and “swiftly” a cognitive agent (human or humanoid) must be able to efficiently integrate multiple streams of information that aid to the learning process itself. Most important among them are social interaction (for example, imitating a teacher’s demonstration), physical interaction (or practice) and “recycling” of previously learnt knowledge (experience) in new contexts. This article presents the skill learning architecture being developed for the humanoid iCub that dynamically integrates multiple streams of learning, multiple task specific constraints and incorporates novel principles that we believe are crucial for constructing a growing motor vocabulary in acting/learning robots. A central feature further is our departure from the well known notion of ‘trajectory formation’ and introduction of the idea of ‘shape’ in the domain of movement. The idea is to learn in an abstract fashion, hence allowing both “task independent” knowledge reuse and task specific “compositionality” to coexist. The scenario of how iCub learns to bimanually coordinate a new tool (a toy crane) to pick up otherwise unreachable objects in its workspace (recycling its past experience of learning to draw) is used to both illustrate central ideas and ask further questions.

Exploratory gaze movements are fundamental for gathering the most relevant information regarding the partner during social interactions. We have designed and implemented a system for dynamic attention allocation which is able to actively control gaze movements during a visual action recognition task. During the observation of a partner’s reaching movement, the robot is able to contextually estimate the goal position of the partner hand and the location in space of the candidate targets, while moving its gaze around with the purpose of optimizing the gathering of information relevant for the task. Experimental results on a simulated environment show that active gaze control provides a relevant advantage with respect to typical passive observation, both in term of estimation precision and of time required for action recognition.

Membrane computing is an interdisciplinary research field focused on new computational models, also known as P systems, inspired by the compartmental model of the cell and the membrane transport mechanisms. Numerical P systems are a type of P systems introduced by Gh. Păun in 2006 for possible applications in economics. Recently, an extension of numerical P systems, enzymatic numerical P systems, has been defined in the context of robot control. This paper presents a new approach to modeling and implementing autonomous mobile robot behaviors and proposes a new odometry module implemented with enzymatic numerical P systems for robot localization. The advantages of modeling robot behaviors with enzymatic membrane controllers and the experimental results obtained on real and simulated robots are also discussed.

This paper is a review of our recent work in which we study insect navigation as a situated and embodied system. This approach has led directly to a novel biomimetic model of route navigation in desert ants. The model is attractive due to the parsimonious algorithm and robust performance. We therefore believe it is an excellent candidate for robotic implementation.

In this paper we formulate basic principles of cognitive human-robot team dynamics following lessons from experimental neuroscience: 1) the cognitive team dynamics in a changing complex environment is transient and can be considered as a temporal sequence of metastable states; 2) the human mental resources –attention and working memory capacity that are available for the processing of sensory and robot generated information in relation to a specific goal– are finite; 3) the interactive cognitive team activity is robust against noise and at the same time sensitive to information from the environment. We suggest a basic dynamical model that describes the evolution of human cognitive and emotion modes and robot information modes together with the dynamics of mental resources. Using this model we have analyzed the team’s dynamical instability, introduced the dynamical description of the information flow capacity, and analyzed the features of the binding dynamics of information flows.

We present an autonomous waiter robot control system based on the reactive layer of the Distributed Adaptive Control (DAC) architecture. The waiterbot has to explore the space where catering is set and invite the guests to serve themselves with chocolate or candies. The control model is taking advantage of DAC’s allostatic control system that allows the selection of actions through the modulation of drive states. In the robot´s control system two independent behavioral loops are implemented serving specific goals: a navigation system to explore the space and a gazing behavior that invites human users to serve themselves. By approaching and gazing at a potential consumer the robot performs its serving behavior. The system was tested in a simulated environment and during a public event where it successfully delivered its wares. From the observed interactions the effect of drive based self-regulated action in living machines is discussed.

We present a neurorobotic model that develops reaching and grasping skills analogous to those displayed by infants during their early developmental stages. The learning process is realized in an incremental manner, taking into account the reflex behaviors initially possessed by infants and the neurophysiological and cognitive maturations occurring during the relevant developmental period. The behavioral skills acquired by the robots closely match those displayed by children. Moreover, the comparison of the results obtained in a control non-incremental experiment demonstrates how the limitations characterizing the initial developmental phase channel the learning process toward better solutions.

Behavioral diversity is one essential feature of living systems in order to exhibit adaptive behavior in hostile and dynamically changing environments. However, classical engineering approaches strive to avoid the behavioral diversity of artificial systems to achieve high performance in specific environments for given tasks. The goals of this research include understanding how living systems exhibit behavioral diversity and use these findings to build robots that exhibit truly adaptive behaviors. To this end, we have focused on an amoeba-like unicellular organism, i.e., the plasmodium of true slime mold. Despite the absence of a central nervous system, the plasmodium exhibits versatile spatiotemporal oscillatory patterns and switches spontaneously between the patterns. Inspired by this, we build a real physical robot that exhibits versatile oscillatory patterns and spontaneous transition between the patterns. The results are expected to shed new light on the design scheme for life-like robots that exhibit amazingly versatile and adaptive behavior.

Neuronal probes can be defined as biohybrid entities where the probes and nerve cells establish a close physical interaction for communicating in one or both directions. During the last decade neuronal probe technology has seen an exploded development. This paper presents newly developed chip–based CyberRat probes for enhanced signal transmission from nerve cells to chip or from chip to nerve cells with an emphasis on

in

−

vivo

interfacing, either in terms of signal−to−noise ratio or of spatiotemporal resolution. The oxide−insulated chips featuring large−scale and high−resolution arrays of stimulation and recording elements are a promising technology for high spatiotemporal resolution biohybrid devices, as recently demonstrated by recordings obtained from hippocampal slices and brain cortex in implanted animals. Finally, we report on SigMate, an

in

−

house

comprehensive automated tool for processing and analysis of acquired signals by such large scale biohybrid devices.

Some species of crayfish can learn and remember environmental features by actively collecting spatial information with their antennae. We were able to qualitatively reproduce features of crayfish exploratory behavior by incorporating into a simulated robot a mapping strategy using cross-correlation of sensed environmental features (wall discontinuities). Our model collects environmental information along one continuous surface (in one dimension) to produce a representation of a two-dimensional space. The simulated robotic model can use this information to discriminate between familiar and novel environments and shows recognition of previously explored environnments. Our results support the hypothesis that crayfish can collect and use spatial information gathered haptically while exploring. Our model also predicts features of this spatial exploration strategy that can be subsequently tested in crayfish, allowing us to modify our model accordingly.

The question of how biological learning and instinctive neural mechanisms interact with each other in the course of development to produce novel, adaptive behaviors was explored via a robotic simulation. Instinctive behavior in the agent was implemented in a hard-wired network which produced obstacle avoidance. Phototactic behavior was produced in two serially connected plastic layers. A self-organizing feature map was combined with a reinforcement learning layer to produce a learning network. The reinforcement came from an internally generated signal. Both the adaptive and fixed networks supplied motor control signals to the robot motors. The sizes of the self-organizing layer, reinforcement layer, and the complexity of the environment were varied and effects on robot phototactic efficiency and accuracy in the mature networks were measured. A significant interaction of the three independent variables was found, supporting the idea that organisms evolve distinct combinations of instinctive and plastic neural mechanisms which are tailored to the demands of the environment in which their species evolved.

This paper presents a computational theory for generating the complicated arm movements needed for tasks such as reaching while avoiding obstacles, or scratching an itch on one arm with the other hand. The required movements are computed using many control units with virtual locations over the entire surface of the arm and hand. These units, called

brytes

, are like little brains, each with its own input and output and its own idea about how its virtual location should move. The paper explains how a previously developed gradient method for dealing with ill-posed multi-joint movements [1] can be applied to large numbers of spatially distributed controllers. Simulations illustrate when the arm movements are successful and when and why they fail. Many of these failures can be avoided by a simple method that adds intermediate reaching goals. The theory is consistent with a number of existing experimental observations.

Faster re–learning of an already acquired motor skill is a common phenomenon observed in classical conditioning of motor responses. The cerebellum is critically involved in the acquisition and retention of those responses. Nevertheless, it remains unclear whether the memory of the association is stored in the cerebellar cortex or in the nuclei. In a previous study we have demonstrated that a neuronal model reflecting basic properties of the cerebellum can acquire well–timed conditioned responses by mediation of plasticity in the cerebellar cortex when controlling a robot in an obstacle avoidance task. Here we extended the model to investigate a possible mechanism of memory transfer and consolidation based on plasticity at the level of the deep nucleus. Experimental results show how the collaboration of these two sites of plasticity can drive the robot to timely adapt and maintain a long term memory in solving the obstacle avoidance task.

In an effort to create an autonomous flying robot, we sought inspiration in the evolution of insects. Previous research on their visual system has led to the creation of navigational models exploiting insect optomotor principles; these are applied in the present study, by using a fast and robust quadrotor. The results show that while a biomimetic approach for an unmanned air vehicle (UAV) is plausible the reliance on vision alone shows instabilities. These can be overcome by using gyroscopes, the technological equivalent of the haltere system found in flying insects.

The first step for building reliable humanoid systems is to provide them with perceptual mechanisms that have human attributes, such as the skill development, social interaction, environmental embodiment and sensorial integration. Despite tactile perception being one of the most important elements for human interaction with the world, its implementation within artificial systems has been tardy, principally because it requires a complete integration with the motor systems and an environmental coupling to extract comprehensible information [1]. Thus, this work aims to generate a platform based on haptic information, allowing humanoids to perceive and represent surrounding objects using concepts fully grounded in the sensorial data.

In this paper a locomotion strategy for a six-limb robot inspired by the octopus is shown. A tight relationship between the muscular system and the nervous systems exists in the octopus. At a high level of abstraction, the same relationship between the mechanical structure and the control of the robot is presented here. The control board sends up to six signals to the limbs, which mechanically perform a stereotypical rhythmical movement. The results show how by coordinating only two limbs an effective locomotion is achieved.

This paper introduces a new designed biomimetic robot frog which has the capability of balancing and controlling the height and length of jumping process. Hence a payload is considered inside the robot to keep its balance in high altitude. The positions of motors which generate the force of jumping operation are evaluated. The obtained results showed significant effect of the new design on jumping, balancing and altitude regulation.

To provide a biological framework to be later applied in robotics, we have devised a delayed comparison task in which subjects discriminate between pairs of vibration delivered either to their whiskers, in rats, or fingertips, in human, with a delay inserted between the two stimuli. The task is to compare two successive stimuli, with different position standard deviations defined by

σ

1

and

σ

2

. By varying the stimulus duration we have observed that rats’ performance improves for longer stimuli, suggesting that for stimuli with a probabilistic structure, evidence can be accumulated over time. On the other hand a change in stimulus duration biased human subjects. This experiment constrains models for the integration of tactile information in robotics.

biochemical information processing and production, supermolecular self-assembly, DNA computing, membrane computing, stochastic pi-calculus.

This paper introduces the idea of WARMOR: Whegs

TM

adaptation and reconfiguration of modular robot with tunable compliance. WARMOR robot revisits the purpose of using Whegs for robust propelling the robot over unknown and scattered terrain by approaching two topics necessary for improved performance: modularity for reconfiguration and tunable compliance for adaptation. The paper describes the robot, especially the design of wheg-derivatives and controller.

The extraordinary manipulation capabilities of the octopus arm generate a big interest in the robotics community. Many researchers have put a big effort in the design, modelling and control of continuum manipulators inspired by the octopus arm. New mathematical tools could be introduced in the robotics community and new sources of inspiration are needed in order to simplify the use of the continuum manipulator. A geometrical exact approach for direct model and a Jacobian method for the inverse model are implemented for a continuum manipulator driven by cables. The first experimental results show an average tip error of less than 6% of the manipulator length, and a good numerical performance to solve the inverse kinetics model.

This paper describes a first prototype of a cephalopod-like biomimetic aquatic robot. The robot replicates the ability of cephalopods to travel in the aquatic environment by means of pulsed jet propulsion. A number of authors have already experimented with pulsed jet thrusting devices in the form of traditional piston-cylinder chambers and oscillating diaphragms. However, in this work the focus is placed in designing a faithful biomimesis of the structural and functional components of the

Octopus vulgaris

, hence the robot is shaped as an exact copy of an octopus and is composed, to a major extent, of soft materials. In addition, the propelling mechanism is driven by a compression/expansion cycle analogous to that found in cephalopods. This work offers a hands-on experience of the swimming biomechanics of chephalopods and an insight into a yet unexplored new mode of aquatic propulsion.

Previous studies of the crayfish have demonstrated an ability to remember enclosure spaces and that this memory is informed by tactile information supplied by the animal’s moving antennae. Simulation and robotic studies form a suitable method for exploring central mechanisms and excluding non-viable alternatives.

Octopus skin was characterized to set design criteria for the artificial skin of an octopus-inspired robot. Young’s moduli, failure strain and ultimate stress were obtained via uniaxial tensile tests. The fracture toughness of the skin was measured by using scissors cutting tests. Silicone rubber is waterproof and has a failure strain higher than 500%, but its fracture toughness is much lower than that of the real octopus skin. To overcome this problem, a knitted nylon fabric was chosen as reinforcement. A fabrication process was developed to optimize the stiffness and improve the quality of the skin artifact. Test results showed that the skin artifact is more flexible and tougher than the real skin.

Presented in this paper are two types of contact sensors developed for an octopus inspired robot. The first type was fabricated using QTC pills. The second type is a soft sensor composed of a thin silicone rubber plate sandwiched between two electrolycra sheets. There is a hole at the centre of the silicone plate. Under contact force, both types of sensor transform from insulators to conductors. Sensitivity tests have been carried out. The soft sensor was implemented into the skin prototype and detected contacts with surfaces, whilst remaining relatively insensitive to lateral or longitudinal extension. Its size and sensitivity can be customized according to the requirement of specific applications.

Biomechanical properties of squid suckers have been studied. The stiffness of the sucker rings was measured. Sucking force at sea level and maximum possible sucking force were obtained by using specially-designed tensile tests. Two biomimetic passive suckers were developed based on the studies of the squid sucker. The first one was a direct copy of the squid sucker while the second one is a much more simplified version, a non-stalked sucker like the octopus suckers. The second design works very effectively and has been implemented in the prototype skin of the octopus arm.

We propose an autonomous decentralized control scheme of an earthworm-like robot for its scaffold-based peristaltic locomotion, on the basis of an analysis using a continuum model. We verify its validity through simulations.

Ophiuroids are a suitable model for understanding the mechanism of resilient animal locomotion, because they can move by self-organizing their arm movements even when the arms are arbitrarily cut off. We observed the locomotion of an ophiuroid that has only one arm, and found that it can move by exploiting subtle terrain irregularities. We modeled this behavior using a simple local reflexive mechanism.

Through interactions with live animals biomimetic robots can be used to analyze social behaviors. We have developed a robotic fish enabling us to examine complex interactions in fish shoals. The system uses small wheeled robots under a water tank. The robots are coupled to a fish replica inside the tank using neodymium magnets. The fish integrate a battery pack and two infrared LEDs that are used to track the replicas in the tank. Here, we describe the procedure to build a fish replica, review the implementation details of our hardware and software and compare it to a previous plotter-based system.

Biomimetics is the development of novel technologies through the distillation of principles from the study of biological systems. It can, in principle, extend to all fields of biological research from physiology and molecular biology to ecology, and from zoology to botany. Another key focus is on complete behaving systems in the form of biomimetic robots. Historically, the term was first used by Otto Schmitt during the 1950s, when he described a biological approach to engineering that he termed biomimetics.

One major challenge in adaptive/developmental robotics is constituted by the need to identify design principles that allow robots to acquire and display different behavioral skills by consistently and scalably integrating new behaviors into their existing behavioral repertoire. In this paper we briefly present a novel method that can address this objective, the theoretical background behind the proposed methodology, and the preliminary results obtained in a series of experiments in which a humanoid iCub robot develops progressively more complex object manipulation skills through an incremental language mediated training process.

The study of natural olfaction can assist in developing more robust and sensitive artificial chemical sensing systems. Here we present the implementation on an indoor fully autonomous wheeled robot of two insect models for odor classification and localization based on moth behavior and the insect’s olfactory pathway. Using the biologically based signal encoding scheme of the Temporal Population Code (TPC) as a model of the antenna lobe, the robot is able to identify and locate the source of odors using real-time chemosensor signals. The results of the tests performed show a successful classification for ethanol and ammonia under controlled conditions. Moreover, a comparison between the results obtained with and without the localization algorithm, shows an effect of the behavior itself on the performance of the classifier, suggesting that the behavior of insects may be optimized for the specific sensor encoding scheme they deploy in odor discrimination.

Autonomous navigation has a high demand on controlling the viewpoint of the visual sensor so that it aims at the object of interest. Ideally, the viewpoint should be maneuvered in a similar way as our human gaze which is driven by either saliency or tasks. We study saliency modelling from image histograms. Our study shows that the visual saliency can be efficiently captured by the traditional 1D image histogram. Experiments over fixational eye tracking data also show that the histogram-based saliency achieves state-of-the-art performance.

The octopus arms totally lack of rigid skeleton, and show unique motor and manipulation capabilities thanks to the skill of varying and controlling the stiffness. To take inspiration for the design of innovative technological actuators for soft robotic systems, we investigated the architecture of the muscle fibers in the octopus arm, and we measured their mechanical performance

in vivo

. The key features “extracted” from the octopus arm have been “translated” into engineering specifications, and the identified requirements have been used to design an artificial muscular hydrostat unit, obtaining an actuating component with controllable stiffness capabilities and various applications for a novel generation of soft-bodied robots.

To perform tasks in human-centric environments, humanoid robots should have the ability to interact with and learn from their environment through the sense of touch. In humans, the loss of this capability can be catastrophic – for instance, the absence of a proprioception sense of limb movement can result in a dramatic loss of the precision and speed of hand movements. Furthermore, not only humans but also animals use tactile sensing to explore their environment, with one notable example being the tapping exploration known as whisking performed by rats with their long facial vibrissae. The movement of tactile sensors against an object surface to generate tactile information is known as active tactile sensing because it relies on actively moving the sensor to generate the tactile sensations. Humans make use of different exploratory procedures (EPs) to extract key information of the objects (e.g. tapping, contour following). It is of great interest how humans and animals develop and select these EPs [6], which motivates the present study into integrating these biomimetic properties within a robotic system.

Intra-body mobility is a promising and challenging development for future surgical robots. This poster presents the bio-inspired features of a novel intra-abdominal robot. This new development is adhesion-reliant and moves against gravity on the inside wall of the human abdomen. The adhesive pads on the robot use a micro-structured surface inspired by tree frogs to obtain wet adhesion. The robot integrates a detachment mechanism inspired by the way geckoes peel off their toes. Locomotion is inspired by amoebas, changing the shape of the robot’s body and alternating adhesion between the moving part of the robot and the part that remains still.

Spiking neural networks implemented using electronic Very Large Scale Integration (VLSI) circuits are promising information processing architectures for carrying out complex cognitive tasks in real-world applications. These circuits are developed using standard silicon technologies, and exploit the analog properties of transistors to emulate the phenomena underlying the computations and the communication in the brain. Neuromorphic multi-neuron systems can provide a low-power and scalable information processing technology, that is optimally suited for advanced and future VLSI processes [1].

Erodium cicutarium L.

plants disperse their seeds by a combination of two dispersal strategies: explosive dispersal, and self-burial dispersal. As the fruits dry, the stresses developed in the structure cause the sudden separation of the seeds that fly away from the plant. Then, once on the ground, the seeds respond to variations in the external humidity : their dispersal unit, is helical when dry and linear when wet. The day-night cycle of humidity results in a coiling and uncoiling motor action, that moves the seed across the surface and into the ground. The present study aims at getting a deeper insight into the self-burial strategy of

Erodium cicutarium

in order to implement this ability into mechanical structures that relying on changes in the external parameters (such as light, temperatures or humidity) can achieve passively the same goal.

Tragopogon dubious

is a small herbaceous plant that uses the wind as dispersal vector for its seeds. The seeds are attached to stalked parachutes which increase the aerodynamic drag force on the seeds. This decreases their rate of descent, and hence increases the total distance traveled. The relatively large natural parachute of

Tragopogon dubious

is an ideal model in a biomimetic structure owing to its relative large size, sturdy and robust structure, and the hierarchical distribution of its fibers. The present contribution describes some preliminary results on the structural properties and aerodynamical behavior of this seed, with the goal of developing new stream of designs of lighter or more robust parachute for possible extra-terrestrial purposes.

Plants represent an amazing source of inspiration for designing and developing adaptive robotic systems, representing plants the best example among living beings of efficient soil exploration. Influence of geometrical and mechanical properties of the living plant roots in the root-soil interaction was investigated in order to fully exploit these biological features in the robotic artefact. The study was performed by means of Finite Element method.

Snakes utilize terrain irregularities and attain propulsion force by pushing their bodies against scaffolds. We have previously proposed a local reflexive mechanism of snake locomotion that exploits its body softness. In this study, we develop a soft-bodied snake-like robot to investigate the validity of the proposed mechanism in the real world.

In this study we explore the manufacture of polymer based, 3D structures for neuronal guidance made by direct laser writing (DLW). This technique enables the optimisation of the support structure via utilising Computer Aided Design and Manufacturing (CAD/CAM). The mechanical and chemical properties can be tuned via custom made polymer constructs. We present results on the polymer synthesis, DLW structuring, neuronal cell culture and guidance which provides a proof-of-concept for these scaffolds for use in peripheral neural biohybrid devices.

The authors aim to develop a biorobotic actuator with a muscle tissue of developing chick embryos driven by a photostimulation. The goal is to find an appropriate method for driving such bio-actuators. To do so, we introduce channelrhodopsin-2 which works as both a photoreceptor and an ion channel. We here report on the deeply interesting experimental results as follows: (1) We employed developing chick embryos, that was electroporated chopWR-venus to prospective limb region. (2) Then, we observed that the muscle tissue actuator is contracted by a blue light stimulation.

This paper introduces a tail optimization method for robotic fish regarding to friction force decreasing. The method is based on Computational Fluid Dynamic (CFD). Hence the hydrodynamic model is derived, its motion directions (up, down, left and right) are analyzed and some simulation studies are done. Then results are presented and the shape and length of tail is optimized to have the minimum friction force in turning to left and right.

So far, we have developed snake-like robots whose locomotion is achieved primarily by ADC (autonomous decentralized control), with only the direction of travel and velocity given by radio control signals [3]. In this paper, we introduce an attempt to improve the navigation system of the robot by using a more intuitive eye wear controller. We will demonstrate its function using a simulation of the robotic system.

In this work we present the design of a new actuated adhesion device inspired by octopus suckers. The octopus suckers are very interesting because they are able to attach in wet conditions on different surfaces, and (as explained in the Kier and Smith hypothesis) the connective tissue fibers of the sucker may store elastic energy, allowing to maintain attachment over extended periods. These features represent a great source of inspiration to conceive innovative adhesion systems working in the same environmental conditions of the biological counterpart. Starting from these premises, we have designed a novel bioinspired adhesion device which exploits the incompressibility of water and a low energy consuming strategy.