论文信息 - Genetic algorithm tuning of connectionist controller for compliant robotic tasks

Genetic algorithm tuning of connectionist controller for compliant robotic tasks

Abstract In this paper, a systematic connectionist controller design approach is proposed to quarantee stability and desired performance of the robotic system for compliant tasks by effectively combining genetic algoritms(GA) with neural classification and learning control techniques. The gains of feedback controller, weighting factors of neural controller, and topology of neural classifier are timed by the GA optimization procedure to achieve good results for force tracking of manipulation robots in contact tasks based on suitable fitness functions. Some compliant robot motion simulation experiments have been performed.

Dusko Katic

[1] Hyungsuck Cho,et al. A neurofuzzy model-based compliance controller with application to a telerobot system , 1996 .

[2] Mo M. Jamshidi,et al. Evolutionary Fuzzy Control of a Flexible-Link , 1997, Intell. Autom. Soft Comput..

[3] Huihe Shao,et al. Genetic evolved neural network for process modelling and optimization , 1996 .

[4] Miomir Vukobratovic,et al. Robot compliance control algorithm based on neural network classification and learning of robot-environment dynamic models , 1997, Proceedings of International Conference on Robotics and Automation.

[5] Yuri Ekalo,et al. New approach to control of robotic manipulators interacting with dynamic environment , 1996, Robotica.

[6] Shuzhi Sam Ge,et al. Genetic algorithm tuning of Lyapunov-based controllers: an application to a single-link flexible robot system , 1996, IEEE Trans. Ind. Electron..