Peer Reviewed Publications
Krister Wolff



List of Abstracts and Sources


TITLE: An Evolutionary Based Approach for Control Programming of Humanoids.

AUTHORS: Krister Wolff and Peter Nordin

YEAR: 2003

SOURCE: In Proceedings of the Third IEEE-RAS International Conference on Humanoid Robots, Humanoids 2003 (pp.). TU München and U Karlsruhe, Germany: Institute of Electrical and Electronics Engineers, Inc.

ABSTRACT: We describe the first instance of a novel approach for control programming of humanoid robots, based on evolution. To overcome some of the difficulties with evolution on real hardware, we use a physically realistic simulation of the robot. The essential idea
is to evolve control programs from first principles on a simulated robot, transfer the programs to the real robot, and continue to evolve on the robot. As the key motivation for using simulators, we describe an on-line learning experiment with a humanoid robot. The Genetic Programming system is implemented as a Virtual Register Machine, with a linear genome, and steady state tournament selection. Evolution created controller programs that made the simulated robot produce forward locomotion behavior. A further application of this system, with two phases of evolution, would be to have a flexible adaptation mechanism that can react to hardware failures in the robot.

FILENAME:
WN_Humanoids03.pdf (244 kB), WN_Humanoids03.ps (4235 kB)



TITLE: Learning Biped Locomotion from First Principles on a Simulated Humanoid Robot using Linear Genetic Programming.

AUTHORS: Krister Wolff and Peter Nordin

YEAR: 2003

SOURCE: In Proceedings of the Genetic and Evolutionary Computation Conference, GECCO 2003 (pp.). Chicago, 12-16 July 2003. Morgan Kaufmann.

ABSTRACT: We describe the first instance of an approach for control programming of humanoid robots, based on evolution as the main
adaptation mechanism. In an attempt to overcome some of the difficulties with evolution on real hardware, we use a physically realistic simulation of the robot. The essential idea in this concept is to evolve control programs from first principles on a simulated robot, transfer the resulting programs to the real robot and continue to evolve on the robot. The Genetic Programming system is implemented as a Virtual Register Machine, with 12 internal work registers and 12 external registers for I/O operations. The individual representation scheme is a linear genome, and the selection method is a steady state tournament algorithm. Evolution created controller programs that made the simulated robot produce forward locomotion behavior. An application of this system with two phases of evolution could be for robots working in hazardous environments, or in applications with remote presence robots.

FILENAME:
WN_gecco03.pdf (196 kB)



TITLE: Walking humanoids for robotics research.

AUTHORS: Krister Wolff and Peter Nordin

YEAR: 2002

SOURCE: In Proceedings of the Second International Workshop on Epigenetic Robotics: Modeling Cognitive Development in Robotic Systems. August 10-11, 2002, Edinburgh, Scottland.

ABSTRACT: We present three humanoid robots aimed as platforms for research in robotics, and cognitive development in robotics systems. The 'priscilla' robot is a 180cm full scale humanoid, and the mid-size prototype is called 'elvis' and is about 70cm tall. The smallest size humanoid is the 'elvina' type, about 28 cm tall. Two instances of 'elvina' have been built to enable experiments with
cooperating humanoids. The underlying ideas and conceptual principles, such as anthropomorphism, embodiment, and mechanisms
for learning and adaptivity are introduced as well.

FILENAME:
WN_EpiRob2002.pdf (106 kB)



TITLE: Sensing and Direction in Locomotion Learning with a Random Morphology Robot.

AUTHORS: Karl Hedman and David Persson and Per Skoglund and Dan Wiklund and Krister Wolff and Peter Nordin

YEAR: 2002

SOURCE: In Proceedings of the Genetic and Evolutionary Computation Conference, GECCO 2002, poster paper (pp.1297). New York, 9-13 July 2002. Morgan Kaufmann.

ABSTRACT: We describe the first instance in sensing and direction with a learning Random Morphology robot. Using GP, it learns to locomote itself in different directions and by letting different solutions master the robot in different situations it can thus follow an arbitrary path.

FILENAME:
HPSWWP_gecco2002.pdf (185 kB)



TITLE: Creation of a Learning, Flying Robot by Means of Evolution.

AUTHORS: Peter Augustsson and Krister Wolff and Peter Nordin

YEAR: 2002

SOURCE: In Proceedings of the Genetic and Evolutionary Computation Conference, GECCO 2002 (pp. 1279-1285). New York, 9-13 July 2002. Morgan Kaufmann.
Awarded "Best Paper in Evolutionary Robotics" at GECCO 2002.

ABSTRACT: We demonstrate the first instance of a real on-line robot learning to develop feasible flying (flapping) behavior, using evolution. Here we present the experiments and results of the first use of evolutionary methods for a flying robot. With nature's own method, evolution, we address the highly non-linear fluid dynamics of flying. The flying robot is constrained in a test bench where timing and movement of wing flapping is evolved to give maximal lifting force. The robot is assembled with standard off-the-shelf R/C servomotors as actuators. The implementation is a conventional steady-state linear evolutionary algorithm.

FILENAME:
AWNGecco2002.pdf (362 kB)



TITLE: Evolution of Efficient Gait with an Autonomous Biped Robot using Visual Feedback.

AUTHORS: Krister Wolff and Peter Nordin

YEAR: 2002

SOURCE: In van Amerongen, J., Jonker, J., Regtien, P., Stramigioli, S., editors, Proceedings of the 8th Mechatronics Forum International Conference 2002 (pp 504-513). June 24-26 2002, University of Twente, Enschede, Netherlands.

ABSTRACT: We have developed an autonomous, walking humanoid robot 'elvina' and performed experiments in evolutionary programming with it, in order to optimize a by hand developed locomotion controller. A steady state evolutionary strategy is running on the robot's onboard computer. Individuals are evaluated and fitness scores are automatically determined using the robots onboard vision system and sensors. By using this system, we evolve gait patterns that locomote the robot in a straighter path and in a more robust way
than the previously manually developed gait did.

FILENAME:
WN_Mechatronics02.pdf (99 kB)



TITLE: A Brute-Force Approach to Automatic Induction of Machine Code on CISC Architectures.

AUTHORS: Felix Kühling and Krister Wolff and Peter Nordin

YEAR: 2002

SOURCE: In Proceedings of the 5th European Conference on Genetic Programming, EuroGP2002. Kinsale, Ireland.

ABSTRACT: The usual approach to address the brittleness of machine code in evolution is to constrain mutation and crossover to ensure syntactic closure. In the novel approach presented here we use no constraints on the operators, they all work blindly on the binaries in memory, but we instead encapsulate the code and trap all resulting exceptions. The new approach presented here for machine code evolution on CISC architectures is based on the observation that modern CPUs can cope with incorrect programmes and report errors to the operating system. This way it is possible to return to very simple genetic operators with the objective of increased performance. Furthermore the instruction set used by evolved programmes is no longer limited by the genetic programming system but only by the CPU it runs on. The mapping between evolution platform and execution plattform becomes almost complete, ensuring correct low-level behaviour of all CPU functions.

FILENAME:
KWN_EuroGP02.pdf (187 kB)



TITLE: Evolving 3d Model Interpretation of Images using Graphics Hardware.

AUTHORS: Fredrik Lindblad and Peter Nordin and Krister Wolff

YEAR: 2002

SOURCE: In Proceedings of the 2002 IEEE Congress on Evolutionary Computation, CEC2002. Honolulu, Hawaii, USA.

ABSTRACT: We present a novel approach for 3d-scene interpretation with numerous applications, for instance in robotics. The models are rendered using 3d graphics hardware and DirectX. Both artificial and real images were used to test the system. More than one target image can be used, allowing stereoscopic vision. These experiments present results of interesting generalization.

FILENAME:
LNW_wcci02.pdf (157 kB)



TITLE: Evolution of Efficient Gait with Humanoids using Visual Feedback.

AUTHORS: Krister Wolff and Peter Nordin.

YEAR: 2001

SOURCE: In Proceedings of the 2nd IEEE-RAS International Conference on Humanoid Robots, Humanoids 2001 (pp. 99-106). Waseda University, Tokyo, Japan: Institute of Electrical and Electronics Engineers, Inc.

ABSTRACT: In this paper we present the autonomous, walking humanoids Priscilla, ELVIS and ELVINA and an experiment using evolutionary adaptive systems. We also present the anthropomorphic principles behind our humanoid project and the multistage development methodology. The adaptive evolutionary system used is a steady state evolutionary strategy running on the robot's onboard computer. Individuals are evaluated and fitness scores are automatically determined using the robots onboard digital cameras and near-infrared range sensor. The experiments are performed in order to optimize a by hand developed locomotion controller. By using this system, we evolved gait patterns that locomote the robot in a straighter path and in a more robust way, than the previously manually developed gait did.

FILENAME:
WN_Humanoids01.pdf (481 kB)



TITLE: Constructing a small humanoid walking robot as a platform for the genetic evolution of walking.

AUTHORS: Jens Ziegler, Krister Wolff, Peter Nordin, and Wolfgang Banzhaf.

YEAR: 2001

SOURCE: In U. Rückert, J. Sitte and U. Witkowski, editors, Proceedings of the 5th International Heinz Nixdorf Symposium: Autonomous Minirobots for Research and Edutainment, AMiRE 2001 (pp. 51-59). Paderborn, Germany: Heinz Nixdorf Institute, University of Paderborn.

ABSTRACT: Walking robots from the next challenge in the field of autonomous robots. This paper describes the construction of a fully autonomous humanoid walking robot as a platform for machine learning algorithms like, e.g., Genetic Programming. Built from off-the-shelf components, the described humanoids are cheap, robust and easy to program, which make them an ideal test platform for several experimental approaches in machine learning, sensor fusion or adaptive control. In addition to these research related topics, the walking robots are an ideal tool for educational purposes.

FILENAME:
ZWNB_AMiRE01.pdf (491 kB)



Home

Standard Disclaimer

Updated: May 28, 2004, Krister Wolff