Beta 1

Title Evolving Robot Gait Using Bio-­Inspired Learning Techniques
Author Nielsen, Thomas
Supervisor Witt, Carsten (Algorithms and Logic, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2010
Abstract In the field of robotics most controls systems are statically programmed for a specific set of tasks. Another way of accomplishing the same tasks is to design self-­‐learning control system, which will obtain a similar solution in time by using bio-­‐inspired learning techniques.The aim of this project is to test two different learning techniques, genetic algorithms and neural networks, and determine which is the best solution for a particular problem. The test case is to train a hexapod to walk by evolving its gait cycle. The training will be performed on both a virtual and a physical robot, which allow for more flexible training scenarios. From this report it is shown that both learning techniques can be used to train the robot gait, and that genetic algorithms has proven to be the best solution for this particular problem.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2010-54
Original PDF ep10_54.pdf (5.43 MB)
Admin Creation date: 2010-09-10    Update date: 2010-09-10    Source: dtu    ID: 266714    Original MXD