Title :
Biped walking pattern generation using reinforcement learning
Author :
Lee, Jungho ; Oh, Jun Ho
Author_Institution :
Korea Adv. Inst. of Sci. & Technol., Daejeon
fDate :
Nov. 29 2007-Dec. 1 2007
Abstract :
In this research, a stable biped walking pattern is generated. The walking pattern is a simple third order polynomial. To find the proper boundary condition, the reinforcement learning algorithm is used. The final velocity of the walking pattern is chosen as learning parameter. To test the algorithm, a simulator that includes the reaction between the foot of the robot and the ground was developed. The algorithm is verified through a simulation.
Keywords :
learning (artificial intelligence); legged locomotion; polynomials; biped walking pattern generation; reinforcement learning; third order polynomial; Boundary conditions; Foot; Humans; Learning; Leg; Legged locomotion; Motion control; Polynomials; Robots; Testing; Biped Walking; CMAC; Reinforcement Learning; Walking Pattern;
Conference_Titel :
Humanoid Robots, 2007 7th IEEE-RAS International Conference on
Conference_Location :
Pittsburgh, PA
Print_ISBN :
978-1-4244-1861-9
Electronic_ISBN :
978-1-4244-1862-6
DOI :
10.1109/ICHR.2007.4813903
