An efficient neural network control for manipulator trajectory tracking with output constraints

• Published in: IEEE ICARM 2017 : 2017 IEEE International Conference on Advanced Robotics and Mechatronics : Hefei & Tai'an, China, August 27-31, 2017 pp. 644 - 649
• Main Authors: Dianye Huang, Chenguang Yang, Wei He, Bin Xu, Chun-Yi Su
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2017
• Subjects: Barrier Lyapunov Function(BLF); constraint; manipulator; Manipulator dynamics; Mechatronics; Neural Network(NN); Neural networks; Trajectory; Uncertainty
• DOI: 10.1109/ICARM.2017.8273238

This paper proposes a trajectory tracking scheme for a constrained manipulator with unknown dynamics is investigated, aiming to track the reference trajectory considering the output state constraints as well as unknown external disturbances. First, a modified tan-type barrier lyapunov function(BLF) is utilized to tackle the effect of constraint. Then, uncertainties are compensated with a radical basis function neural network(RBF-NN), the input number of which is reduce so as to construct a simplified neural network. Besides, boundary theory is also adopted in this paper to eliminate the chattering problem. Finally, the simulation results verify the following three aspects: 1) the constrained controller is able to guarantee the output states subject to the output state constraints; 2) the control scheme with the simplified RBF-NN performs almost the same as the one exactly knows the manipulators dynamics during free space motion; 3) the proposed controller shows robustness in the presence of unknown disturbances.