Distributed Formation Control for Ground Vehicles with Visual Sensing Constraint
Jie Tang · Scholarship at UWindsor (University of Windsor) · 2019
Formation control combined with different tasks enables a group of robots to reach a geographical location, avoid a collision, and simultaneously maintain the designed formation pattern. The connection and perception are critical for a multi-agent formation system, mainly when the robots only use vision as a communication method. However, most visual sensors have limited Field-of-view (FOV), which leaves some blind zones. In this case, a gradient-based distributed control law can be designed to keep every robot in the visible zones of other robots during the formation. This control strategy is designed to be processed independently on each vehicle with no network connection. This thesis assesses the feasibility of applying the gradient descent method to the problem of visual constraint vehicle formation.