No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Embodied Artificial Intelligence, International Seminar, Dagstuhl Castle, Germany, July 7-11, 2003, Revised Papers
... "With the fundamental paradigm shift from a computational to an embodied perspective, the nature of research topics, the theoretical and technical issues, and the disciplines involved have changed dramatically, or in other words, the "landscape" has changed completely." (Pfeifer & Iida 2003:1) In their contribution "Embodied Artificial Intelligence: Trends and Challenges," Pfeifer and Iida [2] point to a paradigm shift from emulating information processes of higher cognition in computer science to an embodied perspective on intelligence through imitation of organic beings. Similarly, this article highlights the implications of the sociology of the body for robotics research and new findings on bodily properties for the technical imitation of embodied beings in environmental relationships. ...
The aim of this contribution is to highlight different aspects of embodied intelligence that add to a broader interdisciplinary perspective on this phenomenon. This article intends to bring together three forms of embodied intelligence from the fields of technology and social research, starting from robotics research that deals with the technical replication of organic bodies, to insights from the sociology of the body outlining the relevance of corporeality in social encounters, and finally to the forms of embodied intelligence in the connection of technical forms with organic bodies and their effects on the perception of self and others within the hybrid research field of "Bodies of Technologies". This tripartite division of embodied intelligence deals with the biophysiological properties of an organism in self-environment relationships (the subject of robotics research), the interaction of a being or system with other beings in self-environment relationships (the subject of the sociology of bodies), and finally, as a third, new level of embodied intelligence, the technically mediated experience of (virtual) self-environment relationships through the somatechnical fusion of bio- and socio-physical properties in human-machine hybrids is presented. Accordingly, the article will name the specificity of each form of embodied intelligence and trace the conditions and peculiarities in the gradual increase of complexity from simple to interactive and finally to virtual bodily intelligence.
... To tackle some of the challenges on the computational level, the interdisciplinary field of embodied AI has started to focus on learning through interaction in a dynamic environment and how this may work in brains and machines (Duan et al., 2021;Iida et al., 2004). ...
The fields of biologically inspired artificial intelligence, neuroscience, and psychology have had exciting influences on each other over the past decades. Especially recently, with the increased popularity and success of artificial neural networks (ANNs), ANNs have enjoyed frequent use as models for brain function. However, there are still many disparities between the implementation, algorithms, and learning environment used for deep learning and those employed by the brain, which is reflected in their differing abilities. I first briefly introduce ANNs and survey the differences and similarities between them and the brain. I then make a case for designing the learning environment of ANNs to be more similar to that in which brains learn, namely by allowing them to actively interact with the world and decreasing the amount of external supervision. To implement this sensorimotor learning in an artificial agent, I use deep reinforcement learning, which I will also briefly introduce and compare to learning in the brain. In the research presented in this dissertation, I focus on testing the hypothesis that the learning environment matters and that learning in an embodied way leads to acquiring different representations of the world. We first tested this on human subjects, comparing spatial knowledge acquisition in virtual reality to learning from an interactive map. The corresponding two publications are complemented by a methods paper describing eye tracking in virtual reality as a helpful tool in this type of research. After demonstrating that subjects do indeed learn different spatial knowledge in the two conditions, we test whether this transfers to artificial agents. Two further publications show that an ANN learning through interaction learns significantly different representations of the sensory input than ANNs that learn without interaction. We also demonstrate that through end-to-end sensorimotor learning, an ANN can learn visually-guided motor control and navigation behavior in a complex 3D maze environment without any external supervision using curiosity as an intrinsic reward signal. The learned representations are sparse, encode meaningful, action-oriented information about the environment, and can perform few-shot object recognition despite not knowing any labeled data beforehand. Overall, I make a case for increasing the realism of the computational tasks ANNs need to solve (largely self-supervised, sensorimotor learning) to improve some of their shortcomings and make them better models of the brain.
... Our low order model intimately couples the neural sensory-motor control to the physical system and its action on the environment, i.e, substrate. This approach is consistent with the theme of 'embodied intelligence' or 'embodiment' [41][42][43][44]. It reflects essential elements in the current understanding of how sea stars control locomotion based on neuroanatomy and behaviour experiments [2, [7][8][9] in the form of a higher-level representations of the neural circuits underlying locomotion as feedback control laws. ...
The oral surface of sea stars is lined with arrays of tube feet that enable them to achieve highly controlled locomotion on various terrains. The activity of the tube feet is orchestrated by a nervous system that is distributed throughout the body without a central brain. How such a distributed nervous system produces a coordinated locomotion is yet to be understood. We develop mathematical models of the biomechanics of the tube feet and the sea star body. In the model, the feet are coupled mechanically through their structural connection to a rigid body. We formulate hierarchical control laws that capture salient features of the sea star nervous system. Namely, at the tube foot level, the power and recovery strokes follow a state-dependent feedback controller. At the system level, a directionality command is communicated through the nervous system to all tube feet. We study the locomotion gaits afforded by this hierarchical control model. We find that these minimally coupled tube feet coordinate to generate robust forward locomotion, reminiscent of the crawling motion of sea stars, on various terrains and for heterogeneous tube feet parameters and initial conditions. Our model also predicts a transition from crawling to bouncing consistently with recent experiments. We conclude by commenting on the implications of these findings for understanding the neuromechanics of sea stars and their potential application to autonomous robotic systems.
... Our low order model intimately couples the neural sensory-motor control to the physical system and its action on the environment, i.e, substrate. This approach is consistent with the theme of "embodied intelligence" or "embodiment" [41][42][43][44]. It reflects essential elements in the current understanding of how sea stars control locomotion based on neuroanatomy and behavior experiments [1,[6][7][8] in the form of a higher level representations of the neural circuits underlying locomotion as feedback control laws. ...
The oral surface of sea stars is lined with arrays of tube feet that enable them to achieve highly controlled locomotion on various terrains. The activity of the tube feet is orchestrated by a nervous system that is distributed throughout the body without a central brain. How such a distributed nervous system produces a coordinated locomotion is yet to be understood. We develop mathematical models of the biomechanics of the tube feet and the sea star body. In the model, the feet are coupled mechanically through their structural connection to a rigid body. We formulate hierarchical control laws that capture salient features of the sea star nervous system. Namely, at the tube foot level, the power and recovery strokes follow a state-dependent feedback controller. At the system level, a directionality command is communicated through the nervous system to all tube feet. We study the locomotion gaits afforded by this hierarchical control model. We find that these minimally-coupled tube feet coordinate to generate robust forward locomotion, reminiscent of the crawling motion of sea stars, on various terrains and for heterogeneous tube feet parameters and initial conditions. Our model also predicts a transition from crawling to bouncing consistent with recent experiments. We conclude by commenting on the implications of these findings for understanding the neuromechanics of sea stars and their potential application to autonomous robotic systems.
... Matarić and Cliff, 1996). 2 Another example to the point is the experiment of Maris and te Boekhorst (1996) that shows how the interaction of agents with an arbitrary neural dynamics -as a matter of fact, Braitenberg vehicles -can lead to a collective, self-organized heap building process. and Nehaniv, 2002), or maximization of some internal reward (Kaplan and Oudeyer, 2004) -that confer a purpose to the self-organization process, and so an explicit raison d'être to the artificial system. ...
... Chentanez et al. (2005) proposed intrinsically motivated reinforcement learning as a framework to allow agents learn hierarchical collections of skills autonomously, and tested it in artificial playrooms. Steels (2004) studied the balance of skill and challenge of behavioral components as the motivation for open ended development of embodied agents. ...
Reinforcement learning for embodied agents is a challenging problem. The accumulated reward to be optimized is often a very rugged function, and gradient methods are impaired by many local optimizers. We demonstrate, in an experimental setting, that incorporating an intrinsic reward can smoothen the optimization landscape while preserving the global optimizers of interest. We show that policy gradient optimization for locomotion in a complex morphology is significantly improved when supplementing the extrinsic reward by an intrinsic reward defined in terms of the mutual information of time consecutive sensor readings.
... pioneer groups can tend towards 'social club(s) for technical elites' (sandvig 2004), attracting members who wish to explore the limits of the technologies (Bina and giaglis 2006b). Frequently there is a sense of playfulness or ludic investigation, engaging with the technologies as an autotelic activity (Csikszentmihalyi 1978;Steel 2004) and a willingness to explore interesting rather than commercially attractive avenues. activists have run audio feeds from local bus stops live onto the internet, mused whether collaborating with graffiti artists might result in spray-on radio antennae that could extend the reach of their networks, and equipped roller skaters with wifi connected video lunchboxes for interviewing local residents. ...
There have been numerous possible scenarios depicted on the impact of the internet on urban spaces. Considering ubiquitous/pervasive computing, mobile, wireless connectivity and the acceptance of the Internet as a non-extraordinary part of our everyday lives mean that physical urban space is augmented, and digital in itself. This poses new problems as well as opportunities to those who have to deal with it. This book explores the intersection and articulation of physical and digital environments and the ways they can extend and reshape a spirit of place. It considers this from three main perspectives: the implications for the public sphere and urban public or semi-public spaces; the implications for community regeneration and empowerment; and the dilemmas and challenges which the augmentation of space implies for urbanists. Grounded with international real -life case studies, this is an up-to-date, interdisciplinary and holistic overview of the relationships between cities, communities and high technologies. © Alessandro Aurigi and Fiorella De Cindio 2008. All rights reserved.
... As the goal of having robots operate in uncontrolled environments becomes more critical to the advancement of robotics, there has been much research on the notion of affordances of objects with respect to a robot agent [1]. Within the context of robotics, affordances describe the possible actions an agent can take acting upon an object and the resulting outcome [2]. Specific examples might include graspable (e.g. ...
A novel behavior representation is introduced that permits a robot to systematically explore the best methods by which to successfully execute an affordance-based behavior for a particular object. The approach decomposes affordance-based behaviors into three components. We first define controllers that specify how to achieve a desired change in object state through changes in the agent's state. For each controller we develop at least one behavior primitive that determines how the controller outputs translate to specific movements of the agent. Additionally we provide multiple perceptual proxies that define the representation of the object that is to be computed as input to the controller during execution. A variety of proxies may be selected for a given controller and a given proxy may provide input for more than one controller. When developing an appropriate affordance-based behavior strategy for a given object, the robot can systematically vary these elements as well as note the impact of additional task variables such as location in the workspace. We demonstrate the approach using a PR2 robot that explores different combinations of controller, behavior primitive, and proxy to perform a push or pull positioning behavior on a selection of household objects, learning which methods best work for each object.
... ijuc_fp(1)_maclennan page 2 2 B. J. MACLENNAN assembly, development, transformation, reconfiguration, or disassembly of the physical system embodying the computation. Embodied computation is based on some of the insights from embodied cognition and embodied artificial intelligence [8,10,11,16,25,26,42,[46][47][48], but extends them to all computation [34,39]. The most common model of computation (binary digital logic) is far removed from the physical processes by which it is implemented, and this has facilitated a beneficial independence of computer design from device technology. ...
Embodied computation is computation in which information processing emerges from and directly governs physical processes. As an example we present artificial morphogenesis, which uses computational processes analogous to those in embryological development in order to assemble complex physical structures. We discuss the requirements for a formalism or programming language for embodied computation oriented toward artificial morphogenesis and present a preliminary design for such a formalism. Use of the formalism is illustrated by several embodied computation problems of increasing complexity.
Theories of decision making are implemented in models that predict and explain behavior in terms of latent cognitive processes. But where do these models come from, and how are they instantiated in the brain? In this article, we examine several avenues where artificial intelligence (AI) and machine learning (ML) can benefit decision theory by providing new methods for developing and testing cognitive models. First, machine learning can be used to efficiently estimate the values of latent parameters in cognitive models and assign posterior probabilities to competing models of the same observed data. Second, models of decision behavior can be embedded within artificially intelligent systems to allow them to make inferences about human counterparts (goals, abilities, cognition) in real time, equipping AI with tools to interact socially. Third, AI can be used to understand how evolutionary and learning processes give rise to the cognitive abilities that support decision making. Last, the tools of experimental psychology and decision sciences can be applied to better understand the “black boxes” of neural networks by systematically testing input–output (stimulus–response) relationships. Put together, we suggest that merging ML/AI into decision-modeling—and vice versa—is a promising path toward many long-term benefits for both fields.
In recent years, bio-inspired robots have shaped numerous domains of technical and scientific production. Bio- inspired robots can now be found in all areas of industry, medicine, architecture, and even culture. Despite the wealth of historiographic and philosophical studies published on this topic, a philosophical investigation of the mimetic principle used in bio-robotics is still missing. In this paper, I will ask a simple question: what is the role of biomimetic and bio-inspired processes in the different practices of bio-robotics?
In the following pages, I will first make some conceptual order. I discuss the differences between several bio- inspired disciplines to clearly identify the biomimetic principle. Second, I introduce the discussion on the neces- sity of imitating nature in early twentieth-century bio-robotics. Third, I state the broader philosophical issue at stake in the debate on the biomimetic principle: the model-world relation, as discussed in the philosophical literature on models and idealization. In section 4, I address several emblematic case studies in which the imitation of nature, variously declined, is fundamental for producing knowledge, thus providing cursory taxonomy of the biomimetic principle. In the conclusion, I will come back to the elephant in the room and suggest how to tackle it further in a fruitful manner. As a broader result, my proposed taxonomy might be used by historians of science and technology as a starting point for historicizing the different practices of current bio-robotics as well as by philosophers to further problematize the various philosophical frameworks that have been accepted and developed in bio-robotics.
The purpose of this study was to investigate how a set of physical and digital instructional activities can serve as an example space to help further develop a concept image that is aligned with the formal concept definition for the limit of a sequence. In addition, the unique affordances and constraints allowed by using either a physical or digital modality in understanding the convergence of a sequence was analyzed. Results suggest that both of these activities served to help students conceptualize the arbitrary nature of the error bound, and for some students it further illustrated the relationship between an arbitrarily small error bound, the limit value, and the index of the sequence. The physical activity constrained students to think of the sequence as a finite terminating set of numbers whereas the digital activity provided additional information that student used in subsequent problem solving.
Achieving greater speeds and densities in the post-Moore’s Law era will require computation to be more like the physical processes by which it is realized. Therefore we explore the essence of computation, that is, what distinguishes computational processes from other physical processes. We consider such issues as the topology of information processing, programmability, and universality. We summarize general characteristics of analog computation, quantum computation, and field computation, in which data is spatially continuous. Computation is conventionally used for information processing, but since the computation governs physical processes, it can also be used as a way of moving matter and energy on a microscopic scale. This provides an approach to programmable matter and programmed assembly of physical structures. We discuss artificial morphogenesis, which uses the formal structure of embryological development to coordinate the behavior of a large number of agents to assemble complex hierarchical structures. We explain that this close correspondence between computational and physical processes is characteristic of embodied computation, in which computation directly exploits physical processes for computation, or for which the physical consequences of computation are the purpose of the computation.
How does human mind develop? What causes developmental disorders? Recent studies suggest the importance of the fetal period in human development. However, study of human fetuses is strongly constrained by technical and ethical difficulties. This project aims at understanding the principles of human development by analyzing and modeling it from the fetal period. Integrating robotics, medicine, psychology, neuroscience, and Tohjisha-kenkyu (first-person view research of developmental disorders), we establish a new trans-disciplinary research field called Constructive Developmental Science. Its contributions include a new understanding of human development and its disorders, comprehensive diagnostic methodologies, and truly appropriate assistive technology.
We define morpho-functional machines as machines which are adaptive not only by means of their neural substrate, but by being able to change their morphology as they perform a task in the real world. Changes in agents can occur at several time scales: short-term, ontogenetic, and phylogenetic. All of these need to be taken into account in designing morpho-functional machines. Studying natural systems, building robots with many degrees of freedom, and artificial evolution (in particular artificial ontogeny) are methods to gain insights into the field. Because the field is only in its infancy, there are no generally accepted methodologies and a multiplicity of methods must be employed to make progress. The paper concludes by identifying a number of important research questions.
We propose a planning algorithm to automatically generate a robust behavior plan (RBP) with which mobile robots can achieve their task goal from any initial states under dynamically changing environments. For this, task description space (TDS) is formulated, where a redundant task configuration space and simulation model of physical space are employed. Successful task episodes are collected, where algorithm is employed. Interesting TDS state vectors are extracted, where occurrence frequency is used. Clusters of TDS state vectors are found by using state transition tuples and features of state transition tuples. From these operations, characteristics of successfully performed tasks by a simulator are abstracted and generalized. Then, a robust behavior plan is constructed as an ordered tree structure, where nodes of the tree are represented by attentive TDS state vector of each cluster. The validity of our method is tested by real robot's experimentation for a box-pushing-into-a-goal task.
The advent of social robots increases significantly the number and the kind of robotics systems to be controlled. Since CSRS
software depends on the particular hardware architecture of a robot, software reuse becomes a significant issue. Reuse is
usually achieved by packaging software in modules, by abstracting a well defined set of functionalities.
Embodied intelligent systems are naturally subject to physical constraints, such as forces and torques (due to gravity and
friction), energy requirements for propulsion, and eventual damage and degeneration. But embodiment implies far more than
just a set of limiting physical constraints; it directly supports the selection and processing of information. Here, we focus
on an emerging link between information and embodiment, that is, on how embodiment actively supports and promotes intelligent
information processing by exploiting the dynamics of the interaction between an embodied system and its environment. In this
light the claim that “intelligence requires a body” means that embodied systems actively induce information structure in sensory
inputs, hence greatly simplifying the major challenge posed by the need to process huge amounts of information in real time.
The structure thus induced crucially depends on the embodied system’s morphology and materials. From this perspective, behavior
informs and shapes cognition as it is the outcome of the dynamic interplay of physical and information theoretic processes,
and not the end result of a control process that can be understood at any single level of analysis. This chapter reviews the
recent literature on embodiment, elaborates some of the underlying principles, and shows how robotic systems can be employed
to characterize and quantify the notion of information structure.
The field of Artificial Intelligence, which started roughly half a cen- tury ago, has a turbulent history. In the 1980s there has been a major paradigm shift towards embodiment. While embodied artificial intelligence is still highly diverse, changing, and far from "theoretically stable", a certain consensus about the important issues and methods has been achieved or is rapidly emerging. In this non-technical paper we briefly characterize the field, summarize its achievements, and identify important issues for future research. One of the fun- damental unresolved problems has been and still is how thinking emerges from an embodied system. Provocatively speaking, the central issue could be ca p- tured by the question "How does walking relate to thinking?"
ResearchGate has not been able to resolve any references for this publication.