Experimental Validation of Multirobot Coverage using Underactuated Networked Mobile Robots

Consequent to previously published numerical results on area coverage control with state estimation, we present apparatus and results of experiments conducted at Bradley University’s robotics and controls laboratories. For that, a team of underactuated wheeled mobile robots is deployed for operating on a two-dimensional (2D) floor area so that its coverage is maximized. Robots are able to communicate with each other through a Wi-Fi network. A remote computer is used to launch the coverage control algorithm to networked robots deployed in the area. It is assumed that each robot is mounted with sensors for measuring the noisy line-of-sight distance between itself and other robots and the noisy angle between its current heading direction and the direction of the other robot positions. Each robot employs a distributed Extended Kalman filter (EKF) to estimate its position and orientation and build its local map by estimating the positions of other robots operating in that area. The robots then partition the area using the conventional Voronoi tessellation technique based on the estimated local map that represents the positions of other robots. Using the conventional Lloyd algorithm, it is guaranteed that the asymptotic convergence of all robots to their corresponding circumcenters of Voronoi cells maximizes the area coverage metric. A set of experimental results with a limited number of robots is presented to evaluate the area coverage metric. To generalize the idea, test runs are also conducted with a large number of robots in the network using the commercially available robot simulator, CoppeliaSim, which mimics the performance as if it would happen in real life.