Contributions published at Artificial Intelligence Lab

The interest in saccadic IOR is funneled by the hypothesis that it serves a clear functional purpose in the selection of fixation points: the facilitation of foraging. In this study, we arrive at a different interpretation of saccadic IOR. First, we find that return saccades are performed much more often than expected from the statistical properties of saccades and saccade pairs. Second, we find that fixation durations before a saccade are modulated by the relative angle of the saccade, but return saccades show no sign of an additional temporal inhibition. Thus, we do not find temporal saccadic inhibition of return. Interestingly, we find that return locations are more salient, according to empirically measured saliency (locations that are fixated by many observers) as well as stimulus dependent saliency (defined by image features), than regular fixation locations. These results and the finding that return saccades increase the match of individual trajectories with a grand total priority map evidences the return saccades being part of a fixation selection strategy that trades off exploration and exploitation.

Recently, the information theoretic approach has been increasingly used in the robotics community as powerful quantitative measures for characterizing the dynamic coupling between the controller, the body, and the environment in embodied robots. This approach is effective and useful even if this interaction regime becomes complex and nonlinear as is often the case in soft robots. In this study, we propose a method for characterizing and visualizing the information transfer spatiotemporally through the robot’s body. This method is based on the framework called “local information transfer” proposed by Lizier et al. We extend it with the permutation-information theoretic approach, which makes it feasible for continuous time series data usually obtained in robotic platforms. To test the power of the proposed method, we performed experiments using a soft robotic arm simulator and a silicone-based soft robotic arm platform inspired by the octopus and showed that the external damage spreading is successfully and clearly visualized by the method. We also analyzed the robustness of the method to noise. Finally, we discuss future applications and possible extensions.

The purpose of this thesis was to investigate the possibility of software development of a new Innovation Management Platform with a tool that uses Business Process Model and Notation (BPMN). Another aim was to compare the \traditional" Object-Oriented (OO) software development approach with the Work ow-based development approach (WF-approach) used by the afore-mentioned BPMN-tool. One of the additional aims was to list and compare the existing Innovation Management Platforms. In order to check the feasibility of the WF-approach for the development of the Innovation Management Platform, a prototype was developed according to the critical software requirements. After the developed prototype was tested, it was gured out that it met the requirements and delivered the main functionality expected from the platform. At the beginning of the development with BPMN-tools, the OO-paradigms were used, which resulted in challenges caused by the conceptual dierence between OO-approach and WF-approach. Nevertheless, after using the WF- paradigms with the BPMN-tools, all discovered conceptual problems were solved. So, the nal software prototype demonstrated that the development with the WF-approach is feasible for developing an online platform for Innovation Management. From the experience and feedback, gained after the source code and work ows of the developed prototype was demonstrated to the reviewers with no technical background, it can be also concluded that software product developed for work ow-based applications with the WF-approach is more self- explanatory than the one developed with the OO-approach, and thus, is easier for the end customers (e.g. business people) to understand.

The biological hypothesis of spinal engine states that locomotion is mainly achieved by the spine, while the legs may serve as assistance. Inspired by this hypothesis, a compliant, multiple degree-of-freedom, biologically-inspired spine has been embedded into a quadruped robot, named Kitty, which has no actuation on the legs. In this paper, we demonstrate how versatile behaviors (bounding, trotting, and turning) can be generated exclusively by the spine's movements through dynamical interaction between the controller, the body, and the environment, known as embodiment. Moreover, we introduce information theoretic approach to quantitatively study the spine internal dynamics and its effect on the bounding gait based on three spinal morphologies. These three morphologies differ in the position of virtual spinal joint where the spine is easier to get bent. The experimental results reveal that locomotion can be enhanced by using the spine featuring a rear virtual spinal joint, which offers more freedom for the rear legs to move forward. In addition, the information theoretic analysis shows that, according to the morphological differences of the spine, the information structure changes. The relationship between the observed behavior of the robot and the corresponding information structure is discussed in detail.

Although the conventional hypothesis which states the coordination of the legs contributes much to locomotion has been widely accepted over the past decades, an alternative one has been proposed with an emphasis on the spine as an engine. In this paper, based on the biological hypothesis of spinal engine, we investigate how morphology of the robot e.g., the choice of actuated joint, the position of rotational joint and the shape and stiffness of the leg, can be adequately exploited to achieve stable and dynamic locomotion. The preliminary experimental results in the real world reveal that the position of rotational joint and shape of the legs are key elements for stable and dynamic locomotion. Based on the results, we discuss the effect of morphology of rear legs, aiming to design a new leg to improve the stability on the spine-driven locomotion.

null

Soft robots have significant advantages over traditional rigid robots because of their morphological flexibility. However, the use of conventional engineering approaches to control soft robots is difficult, especially to achieve autonomous behaviors. With its completely soft body, the octopus has a rich behavioral repertoire, so it is frequently used as a model in building and controlling soft robots. However, the sensorimotor control strategies in some interesting behaviors of the octopus, such as octopus crawling, remain largely unknown. In this study, we review related biological studies on octopus crawling behavior and propose its sensorimotor control strategy. The proposed strategy is implemented with an echo state network on an octopus-inspired, multi-arm crawling robot. We also demonstrate the control strategy in the robot for autonomous direction and speed control. Finally, the implications of this study are discussed.

We develop a bio-inspired controller for an active stereo vision system based on the Hering's law. We extend a model already proposed in literature in two ways. Firstly we evaluate the performance of the controller, inspecting its capability to foveate a generic feature in the 3D space, and the robustness respect to the initial angular configuration of the stereo system. Secondly we introduce the redundant component of the neck. Using a classical learning method we tune the controller to adapt to the controlled system. We investigate how the redundancy is solved by the learned controller, and show that the performance increases and the controlled stereo system generates human-like trajectories.

According to the model of muscle synergies, the central nervous system (CNS) is organised in a modular structure, such that any muscle activation can be produced as a linear superposition of predefined time-varying profiles (i.e. synergies). This organisation might contribute to simplify the control of the musculoskeletal apparatus. Taking inspiration from these findings, we propose a method to identify the synergies that can be used to control a given dynamical system for the task of tracking a set of trajectories. Further, we show how the same approach can be applied to assess the impact of the number of synergies on the performance of the control method. From the theoretical point of view, we provide a novel interpretation of synergies inspired by the Karhunen-Loève decomposition; furthermore, our method suggests that the quality of a set of synergies should be measured in task space rather then in input space.

Soft robots are difficult to control because of their compliant and elastic body dynamics compared with robots made of rigid bodies. In this paper, we present a control scheme inspired by the octopus called timing-based control for soft robotic arms. This control scheme is motivated to positively exploit the natural dynamics of the soft body. We demonstrate a scheme for controlling an object-reaching task by using an echo state network on a 3D physical soft robotic arm simulator and show that this network can successfully perform the task. Detailed analyses and evaluations of the generalization capacity of the network and the performances to the reaching task are presented.

The development of biped machines, inspired by human locomotion, is an interesting subject in engineering science. In order to understand the principles involved in biped locomotion,researchers have proposed several mathematicalframeworks. All these models have provided technical knowledge of biped locomotion that has been applied in the development of many energy efficient biped machines. However, the construction of biped machines capable of exploiting passive dynamics in different gaits remains an unsolved engineering challenge. In this study we propose a controller of the angle of attack that exploits the passive dynamics of a compliant leg to develop stable patterns of locomotion and gait transitions in a defined range of energy. We adopt the spring loaded inverted pendulum (SLIP) model to represent running and walking. The controller naturally emerges from the identification of stable regions of locomotion as well as unstable regions.

Soft robots have significant advantages over traditional robots made of rigid materials. However, controlling this type of robot by conventional approaches is difficult. Reservoir computing has been demonstrated to be an effective approach for achieving rapid learning in benchmark tasks and conventional robots. In this study, we investigated the feasibility and capacity of the reservoir computing approach to embedding and switching between multiple behaviors in a on-line manner in a soft robotic arm. The result shows that this approach can successfully achieve this task.

Sensory information is processed in the dedicated regions of mammalian neural cortex. It is known that neurons in these areas maintain their connections according to the topology of the sensors. Such organization of neurons means that neighboring sensors are mapped into the neighboring neural processing structures. The fundamental characteristic of these sensory areas of the brain is their plasticity. It has been demonstrated that these areas can adapt their structure in response to contingent sensory information. The goal of this project is to analyze tactile feedback information processing mechanisms based on the Hebbian learning. We present two experiments aimed to investigate the basic properties of self-organization in tactile information processing structures. Our experimental setup is based on the designed overlay of 32 sensors assembled on the surface of the robot hand. The first experiment demonstrates the emergence of topologically organized sensory-motor connections induced by the finger-twitching activity of the hand. In the second experiment, we analyze the connectivity learned between different sensors of the hand that were externally stimulated by a rolling ball. Presented results demonstrated that learned connections preserve information about topological organization of sensors in the overlay. It was shown that sensor correlations emerge with respect to the distance and intensity of induced stimulation. Finally, an important relation of observed patterns to the biologically plausible mechanisms of center-surround segregation was discovered.

null

null

Defining the Self is a long-standing quest, which has been addressed by psychologists, mathematicians and philosophers. Prosthetics has become an exciting branch of robotics that carries the potential of answering this question, usinga synthetic or robotic framework, due to the controllability of the relation between the prosthetic device and the human wearer in the interaction with the environment. The incorporation of the robotic prosthesis as part of the wearer’s body has been found to be a sensorimotor artificial transformation subjected to complex technological challenges due to the unstructured environments in which humans operate. This thesis addresses the technological and information-related challenges of haptic interfaces - both haptic sensing and displays - for upper-limb prostheses. It introduces the notion of efficient feedback in prosthetics, a concept through which, technologically, morphology in the design of tactile sensors and haptic displays enhances the relayed information using minimal resources (e.g., electronical, computational and physical). Within the same concept, extended to the information dimension of sensing, this thesis proposes the nature of haptic information which needs to be provided to the prosthesis wearer for acomprehensive environmental representation and an efficient grasp. We show that a quantitative feedback description of proprioceptive sensing,e.g., grip force strength, and exteroceptive sensing, e.g., object slip speed, for prosthetic hands, endows prosthesis users with a robust guidance towards stable grasp, i.e., grip force within safe margins against slip. Additionally, we show the distinct role of grip force and slip speed feedback in regulating the artificial grasp. Following up on these ideas, we developed a haptic device that displays both force and slip in a quantitative way and reveals efficient design principles for prosthetics. We also look at efficient design principles of tactile sensing systems for extracting enhanced haptic information. Ridged patterns on an artificial skin are inspected for their potential to encode haptic stimuli in their morphology during static and dynamic events. We developed a ridged artificial skin that detects stimulus force, slip occurrence, speed and location, by using a single force sensor. Based on evolutionary algorithms, we provide insights into the trade-off between tactile sensing resolution and sensitivity, as an expression of the number and spatial distribution of ridges, respectively. The thesis deepens the understanding of artificial sensorimotor transformations in prosthetic systems and shows the potential of exploiting morphology for efficient sensory feedback schemes in prosthetics.

Complex systems involving many interacting elements often organize into patterns. Two types of pattern formation can be distinguished, static and dynamic. Static pattern formation means that the resulting structure constitutes a thermodynamic equilibrium whose pattern formation can be understood in terms of the minimization of free energy, while dynamic pattern formation indicates that the system is permanently dissipating energy and not in equilibrium. In this paper, we report experimental results showing that the morphology of elements plays a significant role in dynamic pattern formation. We prepared three different shapes of elements (circles, squares, and triangles) floating in a water-filled container, in which each of the shapes has two types: active elements that were capable of self-agitation with vibration motors, and passive elements that were mere floating tiles. The system was purely decentralized: that is, elements interacted locally, and subsequently elicited global patterns in a process called self-organized segregation. We showed that, according to the morphology of the selected elements, a different type of segregation occurs. Also, we quantitatively characterized both the local interaction regime and the resulting global behavior for each type of segregation by means of information theoretic quantities, and showed the difference for each case in detail, while offering speculation on the mechanism causing this phenomenon.

This paper presents a control architecture for redundant and compliant robots inspired by the theory of biological motor primitives which are theorised to be the mechanism employed by the central nervous system in tackling the problem of redundancy in motor control. In our framework, inspired by self-organisational principles, the simulated robot is first perturbed by a form of spontaneous motor activity and the resulting state trajectory is utilised to reduce the control dimensionality using proper orthogonal decomposition. Motor primitives are then computed using a method based on singular value decomposition. Controllers for generating reduced dimensional commands to reach desired equilibrium positions in Cartesian space are then presented. The proposed architecture is successfully tested on a simulation of a compliant redundant robotic pendulum platform that uses antagonistically arranged series-elastic actuation.

The generation of robust periodic movements of complex nonlinear robotic systems is inherently difficult, especially, if parts of the robots are compliant. It has previously been proposed that complex nonlinear features of a robot, similarly as in biological organisms, might possibly facilitate its control. This bold hypothesis, commonly referred to as morphological computation, has recently received some theoretical support by Hauser et al. (Biol Cybern 105:355–370, doi:10.1007/s00422-012-0471-0, 2012). We show in this article that this theoretical support can be extended to cover not only the case of fading memory responses to external signals, but also the essential case of autonomous generation of adaptive periodic patterns, as, e.g., needed for locomotion. The theory predicts that feedback into the morphological computing system is necessary and sufficient for such tasks, for which a fading memory is insufficient. We demonstrate the viability of this theoretical analysis through computer simulations of complex nonlinear mass–spring systems that are trained to generate a large diversity of periodic movements by adapting the weights of a simple linear feedback device. Hence, the results of this article substantially enlarge the theoretically tractable application domain of morphological computation in robotics, and also provide new paradigms for understanding control principles of biological organisms.