Optimum Biped Trajectory Planning for Humanoid Robot Navigation in Unseen Environment

Abstract

Research on humanoid robots in areas related with human-robot interaction has rapidly increased recently especially for application to human living environments and emergency sites. It is apparent that the environments to be shared by humanoid robots are normally dedicated to humans. In this chapter we presented analysis results of optimum biped trajectory planning for humanoid robot navigation to minimize possibility of collision during operation in unseen environment. In this analysis, we utilized 21-dof biped humanoid robot Bonten-Maru II. In this chapter, at first we analyzed the joint trajectory generation in humanoid robot legs to define efficient gait pattern. We present kinematical solutions and optimum gait trajectory patterns for humanoid robot legs. Next, we performed analysis to define efficient walking gait locomotion by improvement of walking speed and travel distance without reducing reduction-ratio at joint-motor system. Next present analyses of collision checking using the robot arms to perform searching, touching and grasping motions in order to recognize its surrounding condition. The presented biped trajectory analysis and planning improved performance of the navigation system. This was proved by simulation and experimental results applying biped humanoid robot Bonten-Maru II during performing biped walk, side-step and yawing motions. Furthermore, analysis of efficient gait trajectory and pattern was presented in this chapter. The analysis results of the gait trajectory generation proposed an efficient gait pattern for the biped robot. Meanwhile, regarding to speed-up walk analysis, simulation results based on humanoid robot Bonten-Maru II parameters revealed that walking speed was improved by applying low duty-ratio at suitable step length and hip-joint height, whereby the walking speed increased about two times compared to normal condition. Moreover, real-time experiments utilizing real biped humanoid robot based on the simulation results showed that the robot's travel distance during walking was improved about three times longer than current walking condition. This analysis results proved that it is possible to improve walking speed in a stable biped locomotion without reducing the reduction-ratio in the robot joint-motor system. Consequently the high torque output at the robot's manipulator to conduct tasks in various motions is maintained. We have presented collision checking method in conjunction with the motion algorithm in the contact sensing-based navigation system. The collision checking is defined by searching motions of the robot's arms that created a radius of detection area within the arm's reach, which is treated as collision free area when no object is detected. Consequently the robot control system created an absolute collision free area for the robot to safely generate trajectories in the navigation tasks. When object is detected during searching motion, the robot arm will touch and grasp the object surface to define self-localization. At this moment, the navigation map is refined and a new path is planned automatically. The above analysis results contribute to the effort to create a stable and reliable biped walking locomotion during performing tasks in the proposed navigation system. Finally, the proposed algorithm was evaluated in an experiment with a prototype humanoid robot operating in a room with walls and obstacles. The experimental results revealed good performance of the robot locomotion in recognizing the environmental conditions and generating suitable locomotion to walk safely towards the target point. Finally, the proposed strategy was demonstrated to have good potential to support current visual-based navigation systems so that humanoids can further `adapt' in real environment.