Cooperative navigation of autonomous vehicles in a known environment

One of the critical goals in the robotics community is to create autonomous robots that interact with unknown environments and will execute high level tasks without human intervention. Developing the technology necessary for autonomous robots is a formidable undertaking with deep interweaved ramifications in automated reasoning, perception and control. It raises many important problems of which motion planning is one of them. Navigation or path planning means to determine how to move an object from its original location and orientation to a goal configuration while avoiding collision with obstacles. The main purpose of this dissertation is to propose cooperative path planning for a group of robots navigating in a known environment. We solve this problem of path planning by introducing a general theory for a set of interconnected Multi Input Multi Output feedback linearizable systems. We show how this theory can be implemented for the path planning of autonomous ground and aerial vehicles. Convergence of a vehicle to a invariant set around the target location is analyzed using the Lyapunov stability method. We propose an obstacle avoidance algorithm by considering the problem as steering the states of a dynamic system to a desired position while avoiding specific regions of state space. Sufficient analytical conditions for avoiding a forbidden region in terms of state trajectories and a set of barrier functions are derived using Lyapunov stability analysis. We then extend the concept of path planning to non feedback linearizable kinematic unicycle models possessing fixed speed and control saturation. The only available control sets the heading of the vehicle. It is shown that a suitable output function allows for convergence of the position states of the unicycle to a bounded neighborhood around a desired set point when the control input is subject to a saturation nonlinearity. Formation control for a group of such vehicles is studied in this dissertation.; We also proposes a possible solution to the problem of tracking mobile targets by group of uninhabited autonomous aerial vehicle (UAV) with communication constraints when only the targets initial position is known. We introduce probabilistic based, time-varying Hospitability maps, which consider the UAV location as points of low hospitability, and extend the Past, Present and Predicted Future (PPP) search algorithm to deal with not only static, but also mobile targets. The practical feasibility of the proposed algorithms are studied by implementing them on Pioneer-3 robots.