EVAL: Evolutionary Computation, Artificial Life & Robotics

Introduction Research Publications People Software, links and media

Introduction

The focus of the Evolutionary Computation and Artificial Life (EVAL) group is the simulated emergence and evolution of complex phenomena. Systems of interest include ecosystems, groups of cells, populations of software agents, robot colonies and even the brain itself. EVAL simulations are typically based on models from the natural and physical sciences combined with the tools of artificial life (AL) and evolutionary computation (EC), such as cellular automata, genetic algorithms and genetic programming.

The brunt of our published work involves simulations of biological systems; the following phenomena have been explored: (a) the migratory movements of fish, birds and polar bears, (b) the evolution of protandry in insects, (c) the evolution of marine microcommunities, and (d) the emergence and evolution of distributed biotic control of the environment (Gaia theory). A rather intense focus on Gaia theory has given us a general interest in the emergence of distributed control systems, both natural and artificial.

We have also begun to explore applications of AL and EC in robotics. Our current focus is on the use of on-board (often incremental) evolutionary algorithms to achieve adaptivity in mobile robots, while our longer-term goal is to investigate similar techniques in colonies of interacting robots.

The following pictures give you a fast visual experience of our activities in the lab. Please follow the links 'Research' and 'Publications' in the menu above and click on the pictures to learn more about our work.

Studies of the interactions of learning and evolution...click to see full-size figure and read more
Reinforced Genetic Programming 		Didabots demo...click to read more       Didabots demo...click to read more      
Swarm Robotics 		Artificial Life and Artificial Chemistries...click to see full-size figure and read more
Gaia Theory & ALife
Embedded evolutionary robotics experiment
Incremental Evolution 		EUZONE simulations...click to read more
Simulations of Microorganisms 		LEGO robot used in the embedded evolutionary robotics experiments
Evolutionary Robotics
Exploring Gaia Theory: Artificial Life on a Planetary Scale...click to read more
Gaian Models 		We use LEGO Mindstorms robotics platform to test our evolutionary robotics experiments on physical robots...click to see the list of links on LEGO Robotics
LEGO Links 		Artificial Life Substrates
ALife Substrates
Vision for Evolutionary Robotics....click to read more
RCXVision 		VISION GUIDED INTELLIGENT TASK PLANNER FOR A ROBOT ARM...click to read more
Replanning

In addition, graduate students in EVAL work on projects in a host of different areas: (a) EAs for predicting stock-market trends, (b) EAs for scheduling in dynamic environments, (c) hybrid EAs that combine evolution and learning, (d) ant-colony optimization applied to communication networks, and (e) computer network security systems based on models of the human immune system.

Although the projects in the EVAL group are varied, we are unified by a general interest in the use of bottom-up, individual-based and evolutionary techniques in both the design of efficient systems and tools, and in the investigation of biological phenomena.

Lab Leader: Professor Keith Downing
Lab Manager: Pavel Petrovic (Ph.D. student)


Norwegian University of Science and Technology  NTNU
Faculty of Information Technology, Mathematics and Electrotechnics  FIME
Department of Computer and Information Science  IDI
 		----
Editor in charge: Keith Downing
Design and maintainance: Pavel Petrovic
Last update: July 2002
 ---- 		NTNU