Using robots to investigate the evolution of adaptive behavior

Bongard, Josh

doi:10.1016/j.cobeha.2015.11.008

Cited by 5 publications

(6 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The essential new feature is the irreducible unity of the controller and the controlled as formulated by the pair of Eqs. [1,2]. The controller is devoid of any system-related functionalities as it is given by a fixed, explicit and contextfree function of the recent history of the sensor values (3).…”

Section: Discussionmentioning

confidence: 99%

“…That this simplicity is an illusion is seen as soon as trying to program a robot for doing a task. While the complexity of programming stands as a challenge for decades, in recent times considerable progress has been achieved by new materials (Kim et al, 2013 ), powerful actuators (Raibert et al, 2008 ), the improved theory of control (Siciliano et al, 2009 ), but in particular by the tremendous increase in computational power that allows modeling and physically realistic simulations of very complex systems to improve planning and control (Mordatch et al, 2012 ; Erez et al, 2013 ; Posa et al, 2014 ) and even allows to simulate large controlled muscular body systems (Yamane and Nakamura, 2011 ), or find new perspectives for artificial evolution (Bongard, 2015 ) by exploiting super computer power. Also there are a variety of new control paradigms around, best demonstrated by the amazing locomotion abilities of the Boston dynamics robots, like BigDog, PETMAN and others.…”

Section: Introductionmentioning

confidence: 99%

“…That this simplicity is an illusion is seen as soon as trying to program a robot for doing a task. While the complexity of programming stands as a challenge for decades, in recent times considerable progress has been achieved by new ma-terials [11], powerful actuators [27], the improved theory of control [30], but in particular by the tremendous increase in computational power that allows modeling and physically realistic simulations of very complex systems to improve planning and control [6,17,24] and even allows to simulate large controlled muscular body systems [35], or find new perspectives for artificial evolution [2] by exploiting super computer power. However, any approach setting on computer power finds its boundaries promptly by the combinatorial complexity of robots in the environment.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Self-Organized Behavior Generation for Musculoskeletal Robots

Der

Martius

2017

Front. Neurorobot.

View full text Add to dashboard Cite

With the accelerated development of robot technologies, control becomes one of the central themes of research. In traditional approaches, the controller, by its internal functionality, finds appropriate actions on the basis of specific objectives for the task at hand. While very successful in many applications, self-organized control schemes seem to be favored in large complex systems with unknown dynamics or which are difficult to model. Reasons are the expected scalability, robustness, and resilience of self-organizing systems. The paper presents a self-learning neurocontroller based on extrinsic differential plasticity introduced recently, applying it to an anthropomorphic musculoskeletal robot arm with attached objects of unknown physical dynamics. The central finding of the paper is the following effect: by the mere feedback through the internal dynamics of the object, the robot is learning to relate each of the objects with a very specific sensorimotor pattern. Specifically, an attached pendulum pilots the arm into a circular motion, a half-filled bottle produces axis oriented shaking behavior, a wheel is getting rotated, and wiping patterns emerge automatically in a table-plus-brush setting. By these object-specific dynamical patterns, the robot may be said to recognize the object's identity, or in other words, it discovers dynamical affordances of objects. Furthermore, when including hand coordinates obtained from a camera, a dedicated hand-eye coordination self-organizes spontaneously. These phenomena are discussed from a specific dynamical system perspective. Central is the dedicated working regime at the border to instability with its potentially infinite reservoir of (limit cycle) attractors “waiting” to be excited. Besides converging toward one of these attractors, variate behavior is also arising from a self-induced attractor morphing driven by the learning rule. We claim that experimental investigations with this anthropomorphic, self-learning robot not only generate interesting and potentially useful behaviors, but may also help to better understand what subjective human muscle feelings are, how they can be rooted in sensorimotor patterns, and how these concepts may feed back on robotics.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…That this simplicity is an illusion is seen as soon as trying to program a robot for doing a task. While the complexity of programming stands as a challenge for decades, in recent times considerable progress has been achieved by new ma-terials [11], powerful actuators [27], the improved theory of control [30], but in particular by the tremendous increase in computational power that allows modeling and physically realistic simulations of very complex systems to improve planning and control [6,17,24] and even allows to simulate large controlled muscular body systems [35], or find new perspectives for artificial evolution [2] by exploiting super computer power. However, any approach setting on computer power finds its boundaries promptly by the combinatorial complexity of robots in the environment.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Self-Organized Behavior Generation for Musculoskeletal Robots

Der

Martius

2017

Front. Neurorobot.

View full text Add to dashboard Cite

show abstract

“…In contrast, there are humanoid robots which can replicate typical human characteristics in the form of physical appearance, movements, words, facial expressions, etc. An example system in which robot's responses evolved by means of an adaptive behavior is presented in [12].…”

Section: Introductionmentioning

confidence: 99%

Proactive Robots With the Perception of Nonverbal Human Behavior: A Review

2019

View full text Add to dashboard Cite

Intelligent robot companions contribute significantly in improving living standards in the modern society. Therefore, human-like decision making skills are sought after during the design of such robots. On the one hand, such features enable the robot to be easily handled by its non-expert human user. On the other hand, the robot will have the capability of dealing with humans without causing any disturbance by the robot's behavior. Mimicing human emotional intelligence is one of the best and reasonable ways of laying the foundation for robotic emotional intelligence. As robots are widely deployed in social environments, perception of the situation or intentions of a user prior to an interaction is required to be proactive. Proactive robots are required to understand what is communicated by the human body language prior to approaching a human. Social constraints in an interaction could be demolished by this assessment in this regard. In this review, we incorporate various findings of human-robot interaction, social robotics and psychophysiology to assess intelligent systems which were capable of evaluating the emotional state of humans prior to an interaction. Second, we identify the cues and evaluation techniques that were utilized by such intelligent agents to simulate and evaluate the suitability of a proactive interaction. Available literature has been evaluated to distinguish limitations of existing methods and suggest possible improvements. These limitations, guiding principles to be adhered to and suggested improvements, are presented as an outcome of the review. INDEX TERMSInteraction initiation, nonverbal cues, context-awareness, social robots, human-robot interaction. A. G. BUDDHIKA P. JAYASEKARA (M'07-SM'17) received the B.Sc. and M.Sc. degrees in electrical engineering from the University of Moratuwa, respectively, in 2004 and 2006, and the doctoral degree in robotics and intelligent systems from Saga University, Japan. He received the Japanese Government Monbukagakusho Scholarship to pursue postgraduate studies in Japan. He joined the Department of Electrical Engineering, University of Moratuwa, after completing his doctoral studies in 2010, where he currently a Senior Lecturer.His main research interests include artificial intelligence techniques, human-robot interaction, human-friendly robotics, machine learning, and intelligent systems. He is the Principle Investigator of the Intelligent Service Robotics Group, Department of Electrical Engineering, University of Moratuwa, and working as the Director of Center for Advanced Robotics (CAR), a multidisciplinary research center of University of Moratuwa from 2017. Currently, he is working on two main projects: intelligent mobile robot for domestic environments (MIRob-Moratuwa Intelligent Robot) and intelligent wheelchair for domestic environments. D. P. CHANDIMA (M'03-SM'17) received the B.Sc. degree from the

show abstract

“…To date, efforts to model and test hypotheses about the evolution of modularity have focused on using non-embodied systems to test ideas drawn from genetic regulatory networks and artificial neural networks (ANNs) (Voordeckers et al, 2015). However, some work has been dedicated to the specific challenges of evolving modularity in embodied systems (Bongard, 2011(Bongard, , 2015Bernatskiy and Bongard, 2015;Bongard et al, 2015). Contrary to the prediction that modularity evolves in response to selection on performance alone, it has been shown to evolve, instead, as a by-product of selection for enhanced performance and reduced connection costs (Clune et al, 2013).…”

mentioning

confidence: 99%

Modularity and Sparsity: Evolution of Neural Net Controllers in Physically Embodied Robots

et al. 2016

View full text Add to dashboard Cite

While modularity is thought to be central for the evolution of complexity and evolvability, it remains unclear how systems bootstrap themselves into modularity from random or fully integrated starting conditions. Clune et al. (2013) suggested that a positive correlation between sparsity and modularity is the prime cause of this transition. We sought to test the generality of this modularity-sparsity hypothesis by testing it for the first time in physically embodied robots. A population of 10 Tadros -autonomous, surface-swimming robots propelled by a flapping tail -was used. Individuals varied only in the structure of their neural net controller, a 2 × 6 × 2 network with recurrence in the hidden layer. Each of the 60 possible connections was coded in the genome and could achieve one of three states: −1, 0, and 1. Inputs were two light-dependent resistors and outputs were two motor control variables to the flapping tail, one for the frequency of the flapping and the other for the turning offset. Each Tadro was tested separately in a circular tank lit by a single overhead light source. Fitness was the amount of light gathered by a vertically oriented sensor that was disconnected from the controller net. Reproduction was asexual, with the top performer cloned and then all individuals entered into a roulette wheel selection process, with genomes mutated to create the offspring. The starting population of networks was randomly generated. Over 10 generations, the population's mean fitness increased twofold. This evolution occurred in spite of an unintentional integer overflow problem in recurrent nodes in the hidden layer that caused outputs to oscillate. Our investigation of the oscillatory behavior showed that the mutual information of inputs and outputs was sufficient for the reactive behaviors observed. While we had predicted that both modularity and sparsity would follow the same trend as fitness, neither did so. Instead, selection gradients within each generation showed that selection directly targeted sparsity of the connections to the motor outputs. Modularity, while not directly targeted, was correlated with sparsity, and hence was an indirect target of selection, its evolution a "by-product" of its correlation with sparsity.

show abstract

Using robots to investigate the evolution of adaptive behavior

Cited by 5 publications

References 46 publications

Self-Organized Behavior Generation for Musculoskeletal Robots

Self-Organized Behavior Generation for Musculoskeletal Robots

Proactive Robots With the Perception of Nonverbal Human Behavior: A Review

Modularity and Sparsity: Evolution of Neural Net Controllers in Physically Embodied Robots

Contact Info

Product

Resources

About