FPGA-based visual control system using dynamic perceptibility

Abstract

This paper presents an embedded architecture based on field programmable gate arrays technology to implement direct visual controllers. These controllers directly generate the joint torques to perform visual servoing tasks taking into account the robot dynamics. Additionally, the concept of dynamic perceptibility, which provides information about the system capability to track objects using direct visual servoing, is defined. This concept is integrated in the proposed controllers and implemented in the embedded architecture for robot and image singularity avoidance. To do this, the paper presents an optimal control framework for direct visual servoing robots which integrates the concept of dynamic perceptibility. This paper presents the formulation of the optimal framework and details the implementation of one of the derived controllers in the FPGA-based architecture. In order to evaluate the proposed architecture, the FPGA architecture is applied to two different robots. The first one is the industrial robot Mitsubishi PA10 with 7 degrees of freedom and the second one is a 4 degrees of freedom robot designed by the authors. Additionally, different experiments are described to show the system behavior when a trajectory close to a singularity is obtained.