Real-Time Trajectory Generation for Both Arms of a Humanoid Robot

This paper proposes a real-time trajectory generation algorithm for both arms of a humanoid robot. Since it is hard to find a closed form of inverse kinematics for each arm of seven degrees of freedom, the damped least-squares method is employed to obtain the inverse kinematics. The trajectory is generated by the minimum-jerk method to maximize the position accuracy. Considering the performance in computation time, a software SD/FAST is used to find a Jacobian matrix of the arm. Computer simulation was performed to verify the effectiveness of the proposed algorithm using a Webot simulator for the upper body of Mybot developed in the RIT Lab., KAIST. The results show that the proposed algorithm generates trajectory in real-time and it is robust to singularity.