Evolutionary Robotics (ER) is a methodology that uses evolutionary computation to develop controllers for autonomous robots. Algorithms in ER frequently operate on populations of candidate controllers, initially selected from some distribution. This population is then repeatedly modified according to a fitness function. In the case of genetic algorithms (or "GAs"), a common method in evolutionary computation, the population of candidate controllers is repeatedly grown according to crossover, mutation and other GA operators and then culled according to the fitness function. The candidate controllers used in ER applications may be drawn from some subset of the set of artificial neural networks, although some applications (including SAMUEL, developed at the Naval Center for Applied Research in Artificial Intelligence) use collections of "IF THEN ELSE" rules as the constituent parts of an individual controller. It is theoretically possible to use any set of symbolic formulations of a control laws (sometimes called a policies in the machine learning community) as the space of possible candidate controllers. It is worth noting that artificial neural networks can also be used for robot learning outside of the context of evolutionary robotics. In particular, other forms of reinforcement learning can be used for learning robot controllers. In computer science evolutionary computation is a subfield of artificial intelligence (more particularly computational intelligence) involving combinatorial optimization problems. ... Basic Principles A controller is the brain component of a system that monitors certain input variables and adjusts other output variables to achieve the desired operation. ... Autonomous robots are robots which can perform desired tasks in unstructured environments without continuous human guidance. ... Flowcharts are often used to represent algorithms. ... Basic Principles A controller is the brain component of a system that monitors certain input variables and adjusts other output variables to achieve the desired operation. ... In mathematics and statistics, a probability distribution, more properly called a probability density, assigns to every interval of the real numbers a probability, so that the probability axioms are satisfied. ... A fitness function is a particular type of objective function that quantifies the optimality of a solution (that is, a chromosome) in a genetic algorithm so that that particular chromosome may be ranked against all the other chromosomes. ... A genetic algorithm (or short GA) is a search technique used in computing to find true or approximate solutions to optimization and search problems. ... In computer science evolutionary computation is a subfield of artificial intelligence (more particularly computational intelligence) involving combinatorial optimization problems. ... A genetic algorithm (or short GA) is a search technique used in computing to find true or approximate solutions to optimization and search problems. ... A fitness function is a particular type of objective function that quantifies the optimality of a solution (that is, a chromosome) in a genetic algorithm so that that particular chromosome may be ranked against all the other chromosomes. ... An artificial neural network (ANN) or commonly just neural network (NN) is an interconnected group of artificial neurons that uses a mathematical model or computational model for information processing based on a connectionist approach to computation. ... Basic Principles A controller is the brain component of a system that monitors certain input variables and adjusts other output variables to achieve the desired operation. ... Look up policy in Wiktionary, the free dictionary. ... As a broad subfield of artificial intelligence, Machine learning is concerned with the development of algorithms and techniques that allow computers to learn. At a general level, there are two types of learning: inductive, and deductive. ... An artificial neural network (ANN) or commonly just neural network (NN) is an interconnected group of artificial neurons that uses a mathematical model or computational model for information processing based on a connectionist approach to computation. ... Robot learning is a subset of machine learning and robotics. ... Reinforcement learning refers to a class of problems in machine learning which postulate an agent exploring an environment in which the agent perceives its current state and takes actions. ... Basic Principles A controller is the brain component of a system that monitors certain input variables and adjusts other output variables to achieve the desired operation. ...

History

The foundation of ER was laid with work at the national research council in Rome in the 90s, but the initial idea of encoding a robot control system into a genome and have artificial evolution improve on it dates back to the late 80s. An evolutionary algorithm (also EA, evolutionary computation, artificial evolution) is a generic term used to indicate any population-based optimization algorithm that uses mechanisms inspired by biological evolution, such as reproduction, mutation and recombination (see genetic operators). ...

The term evolutionary robotics was introduced in 1993 by Cliff, Harvey and Husbands at the University of Sussex. In 1992 and 1993 two teams, a team surrounding Floreano and Mondada at the EPFL in Lausanne and a research group at the COGS at the University of Sussex reported the first experiments on artificial evolution of autonomous robots. The success of this early research triggered a wave of activity in labs around the world trying to harness the potential of the approach. Dave Cliff (25 June 1944– ) is a British jazz musician who plays guitar. ... Inman Harvey is a Senior Lecturer in CSAI at the University of Sussex [1]. His research interests largely centre on the development of artificial evolution as an approach to the design of complex systems. ... Phil Husbands is a professor of Computer Science and Artificial Intelligence at the University of Sussex [1]. His research interests are in Long-term investigation of artificial evolution of nervous systems for robots, with emphasis on visually guided robots acting in the real world; theoretical and practical development of advanced... The University of Sussex is an English campus university located near the East Sussex village of Falmer, near Brighton and Hove. ... Dario Floreano is the professor of the Laboratory of Intelligent Systems (LIS) [1] of the École Polytechnique Fédérale de Lausanne in Switzerland. ... Francesco Mondada is a Swiss researcher in artificial intelligence. ... The Monster Clothespin from Outer Space, and entrance of the EPFL The École Polytechnique Fédérale de Lausanne (EPFL) is the Swiss Federal Institute of Technology in Lausanne in Switzerland. ... Lausanne is a city in the French-speaking part of Switzerland, situated on the shores of Lake Geneva (French: Lac Léman), and facing Évian-les-Bains (France) and with the Jura hills to its north. ... The University of Sussex is an English campus university located near the East Sussex village of Falmer, near Brighton and Hove. ...

Lately, the difficulty in "scaling up" the complexity of the robot tasks has shifted attention somewhat towards the theoretical end of the field rather than the engineering end.

Evolutionary Robotics

Evolutionary robotics is done with many different objectives, often at the same time. These include creating useful controllers for real-world robot tasks, exploring the intricacies of evolutionary theory (such as the Baldwin effect), reproducing psychological phenomena, and finding out about biological neural networks by studying artificial ones. Creating controllers via artificial evolution requires a large number of evaluations of a large population. This is very time consuming, which is one of the reasons why controller evolution is usually done in software. Also, initial random controllers may exhibit potentially harmful behaviour, such as repeatedly crashing into a wall, which may damage the robot. Transferring controllers evolved in simulation to physical robots is very difficult and a major challenge in using the ER approach. The reason is that evolution is free to explore all possibilities to obtain a high fitness, including any inaccuracies of the simulation. This need for a large number of evaluations, requiring fast yet accurate computer simulations, is one of the limiting factors of the ER approach. The Baldwin effect is a theory of James Mark Baldwin in which individual learning of a characteristic significantly affects the evolution of a species with respect to that characteristic. ...

In rare cases, evolutionary computation may be used to design the physical structure of the robot, in addition to the controller. One of the most notable examples of this was Karl Sims' demo for Thinking Machines corporation. Karl Sims is a researcher formerly with the MIT Media Lab who is most well known for using genetic programming to evolve virtual creatures that competed in various simulated environments as described in this paper. ... Thinking Machines Corporation was a supercomputer manufacturer founded in Waltham, Massachusetts in 1982 by W. Daniel Hillis and Sheryl Handler to turn Hilliss doctoral work at MIT on massively parallel computing architectures into a commercial product called the Connection Machine. ...

Motivation for Evolutionary Robotics

Many of the commonly used machine learning algorithms require a set of training examples consisting of both a hypothetical input and a desired answer. In many robot learning applications the desired answer is an action for the robot to take. These actions are usually not known explicitly a priori, instead the robot can, at best, receive a value indicating the success or failure of a given action taken. Evolutionary algorithms are natural solutions to this sort of problem framework, as the fitness function need only encode the success or failure of a given controller, rather than the precise actions the controller should have taken. An alternative to the use of evolutionary computation in robot learning is the use of other forms of reinforcement learning, such as q-learning, to learn the fitness of any particular action, and then use predicted fitness values indirectly to create a controller. As a broad subfield of artificial intelligence, Machine learning is concerned with the development of algorithms and techniques that allow computers to learn. At a general level, there are two types of learning: inductive, and deductive. ... A training set is used in artificial intelligence, together with a supervised training method, and it consists of an input vector and an answer vector. ... Robot learning is a subset of machine learning and robotics. ... ASIMO, a humanoid robot manufactured by Honda. ... Robot learning is a subset of machine learning and robotics. ... Reinforcement learning refers to a class of problems in machine learning which postulate an agent exploring an environment in which the agent perceives its current state and takes actions. ... Q-learning is a reinforcement learning technique that works by learning an action-value function that gives the expected utility of taking a given action in a given state. ...

Links and further Information

• An introduction to Evolutionary Robotics with annotated bibliography

• A gentle introduction to ER

See also

• Evolutionary computation
• Roboticist
• Robotics
• Cognitive robotics
• Artificial Life
• Artificial intelligence
• Cybernetics

In computer science evolutionary computation is a subfield of artificial intelligence (more particularly computational intelligence) involving combinatorial optimization problems. ... Definition A roboticist conceptualizes, designs, builds, programs, and experiments with robots. ... It has been suggested that this article or section be merged with robot. ... Cognitive robotics (CR) is concerned with endowing robots (which have to operate in complex, fast-changing environments) with high-level cognitive capabilities, such as anticipation, planning, reasoning about other agents, and reasoning about their own mental states. ... Artificial life, also known as alife or a-life, is the study of life through the use of human-made analogs of living systems. ... Hondas humanoid robot AI redirects here. ... Cybernetics is the study of communication and control, typically involving regulatory feedback in living organisms, machines and organisations, as well as their combinations. ...

Web

Academic institutes and researchers

• University of Sussex: Inman Harvey, Phil Husbands, Ezequiel Di Paolo, Eric Vaughan, Thomas Buehrmann
• CNR: Stefano Nolfi, Raffaele Calabretta
• EPFL: Dario Floreano
• University of Zürich: Rolf Pfeifer
• Cornell University: Hod Lipson, Josh Bongard
• Case Western Reserve University: Randall Beer
• Center for Robotics and Intelligent Machines, NCSU: Eddie Grant, Andrew Nelson
• University College London: Peter Bentley, Siavash Haroun Mahdavi
• University of Essex: Simon Lucas
• Brandeis University: Jordan Pollack
• University of Skövde: Tom Ziemke
• U.S. Naval Research Laboratory's, Navy Center for Applied Research In Artificial Intelligence: Alan C. Schultz, Mitchell A. Potter, Kenneth De Jong
• Fraunhofer AiS, Intelligent Dynamics Dep.: Frank Pasemann
• Evolved Virtual Creatures by Karl Sims (GenArts)
• [1] Ken Rinaldo artificial life robotics

The University of Sussex is an English campus university located near the East Sussex village of Falmer, near Brighton and Hove. ... Inman Harvey is a Senior Lecturer in CSAI at the University of Sussex [1]. His research interests largely centre on the development of artificial evolution as an approach to the design of complex systems. ... Phil Husbands is a professor of Computer Science and Artificial Intelligence at the University of Sussex [1]. His research interests are in Long-term investigation of artificial evolution of nervous systems for robots, with emphasis on visually guided robots acting in the real world; theoretical and practical development of advanced... Ezequiel Di Paolo is a Lecturer in Evolutionary and Adaptive Systems at the University of Sussex [1]. His research interests are in evolutionary robotics studying lifetime adaptation to radical sensorimotor disruptions. ... Eric Vaughan is a researcher at the University of Sussex [1]. His research involves combining neural networks, evolutionary algorithms, and passive dynamics to evolve the bodies and controls systems of bipedal robots. ... The Italian National Research Council (CNR) is a public organization of great relevance in the field of scientific and technological research of the Country whose original institution goes back to year 1923. ... Dr. Stefano Nolfi is a senior researcher of the of the Institute of Codnilation Reforms and Society, CNR and head of the Laboratory of Autonomous Robots and has a title of genious. ... The Monster Clothespin from Outer Space, and entrance of the EPFL The École Polytechnique Fédérale de Lausanne (EPFL) is the Swiss Federal Institute of Technology in Lausanne in Switzerland. ... Dario Floreano is the professor of the Laboratory of Intelligent Systems (LIS) [1] of the École Polytechnique Fédérale de Lausanne in Switzerland. ... The University of Zurich (in German: Universität Zürich) is the largest university of Switzerland, in the city of Zurich. ... Rolf Pfeifer received his masters degree in physics and mathematics and his Ph. ... Cornell redirects here. ... Josh Bongard received his Bachelors degree in Computer Science from McMaster University, Canada, his Masters degree from the University of Sussex, UK, and his PhD from the University of Zurich, Switzerland. ... Case Western Reserve University is a private research university located in Cleveland, Ohio, United States, with some dormitories on the south end of campus located in Cleveland Heights. ... Randall D. Beer is a professor of electrical engineering and computer science, professor of biology, professor of cognitive science, and director of the Dynamics of Adaptive Behavior research group at Case Western Reserve University. ... North Carolina State University Seal North Carolina State University is an institution of higher learning located in Raleigh, North Carolina. ... Eddie Grant is the name of the following people: Eddie Grant (musician) - a Black British musician Eddie Grant (baseball player) This is a disambiguation page — a navigational aid which lists other pages that might otherwise share the same title. ... University College London, commonly known as UCL, is one of the colleges that make up the University of London. ... Peter John Gerald Bentley (born March 17, 1930) is a Canadian businessman and the third Chancellor of the University of Northern British Columbia. ... The University of Essex is a British plate glass university. ... Brandeis University is a private university in Waltham, Massachusetts, United States. ... The US Naval Research Laboratory (NRL) is the corporate research laboratory for the United States Navy and the United States Marine Corps and conducts a broad program of scientific research and advanced development. ... Karl Sims is a researcher formerly with the MIT Media Lab who is most well known for using genetic programming to evolve virtual creatures that competed in various simulated environments as described in this paper. ...

Books

• Evolutionary Robotics by Stefano Nolfi and Dario Floreano. ISBN 0-262-14070-5
• Advances in the Evolutionary Synthesis of Intelligent Agents by Mukesh Patel, Vasant Honavar and Karthik Balakrishnan (Ed). Cambridge, MA: MIT Press. 2001. ISBN 0-262-16201-6

Dr. Stefano Nolfi is a senior researcher of the of the Institute of Codnilation Reforms and Society, CNR and head of the Laboratory of Autonomous Robots and has a title of genious. ... Dario Floreano is the professor of the Laboratory of Intelligent Systems (LIS) [1] of the École Polytechnique Fédérale de Lausanne in Switzerland. ... Professor Vasant Honavar is an American computer scientist, specializing in artificial intelligence. ...

Main Conferences

• Evolutionary Robotics
• GECCO
• IEEE Congress on Evolutionary Computation
• European Conference on Artificial Life
• ALife