Paper Info
Title
Evolving Humanoid Robot Motions Based on Gene Regulatory Network
Abstract
Motion generation for humanoid robots is a hard task because of the complex and unstable structures. We solve this task using MONGERN which is a method to generate robot motions based on GRN with an evolutionary approach. MON GERN does not require any calculation of dynamic properties but can generate suitable motions for robots. In this research, we tried to obtain three realistic motions in the simulator called Webots to show the robustness and effectiveness of MONGERN. The motions are trap motion in soccer, bracing motion for a predictable impact and continuous skating motion. In our experiments, we were able to obtain desirable motions which include unexpected motions. We were able to observe the robust and flexible motion-generation ability of this method. In the case of skating motion, it has been successfully applied to a real robot, KHR3HV designed by Kondo-Science. These experiments imply that the method can be useful in generating complex motions even in the real world.
Year
DOI
Venue
2019
10.1109/ICARM.2019.8834061
2019 IEEE 4th International Conference on Advanced Robotics and Mechatronics (ICARM)
Keywords
Field
DocType
gene regulatory network,motion generation,MONGERN,evolutionary approach,trap motion,continuous skating motion,humanoid robot motions,Webots,bracing motion
Simulation,Computer science,Motion generation,Robustness (computer science),Robot,Gene regulatory network,Humanoid robot
Conference
ISBN
Citations 
PageRank 
978-1-7281-0065-4
0
0.34
References 
Authors
7
2
Name
Order
Citations
PageRank
Naoyuki Kawabata100.34
Hitoshi Iba21541138.51