Cooperative Navigation among Multiple Mobile Robots

Abstract

A group of mobile robots equipped with interrobot position sensors and intercommunication of control information can improve their ability to follow prescribed trajectories using a cooperative navigation system (CNS). In the formulation of this approach, the interrobot positions are treated as measurements in a Kalman filter which estimates the positions of all N robots simultaneously. Analysis and simulation of the CNS algorithm under the linear measurement model with Gaussian error distributions, and no additional external sensors, shows three main properties: (1) The mean-square error of each robot is always improved by the CNS system. (2) The mean-square error of each robot decreases as the number of robots, N, is increased. (3) Without external sensors, the mean-square error of each robot increases monotonically along the path. With the addition of external sensors to any one of the robots, the CNS algorithm intrinsically propagates the sensor information throughout the group. The CNS approach described here may be useful in applications such as construction, maintenance tasks, and exploration of domains, including space and underwater, where groups of cooperating robots require coordination of trajectories and avoidance of conflicts and collision.