Real-time visual predictive control of manipulation systems

摘要

This paper presents the results obtained by visual predictive control laws in real-time visual servoing architectures. A key role when designing a stable image based control law is played by the visual features, and thus both points and image moments were considered and their performance revealed. A 6 d.o.f manipulator robot with an eye-in-configuration is used to design a visual servoing architecture and realtime implementation using Matlab is described. The results of different experiments for planar static objects are presented and discussed. For establishing the performance of visual predictive control laws a comparison with classical image based control laws is conducted.