Landmarks based path planning for UAVs in GPS-denied areas∗

In this paper, we propose a UAV path planner travelling from a given source to goal location in GPS-denied areas. The environment consists of a set of cellular towers which are treated as landmarks having finite communication range. The vehicle has to perform dead reckoning in regions where landmarks are not available. Therefore, the objective is to determine a time optimal path taking the presence of landmarks into account while ensuring the covariance due to dead-reckoning is within a given bound. Solving the stochastic optimal control problem to determine a path in the continuous domain is very difficult and hence we discrete the path as a set of way points and optimize the location of these way points to obtain a time optimal path satisfying the covariance bounds. We use a particle swarm optimization technique coupled with a rabbit-carrot based path following technique to determine a near-optimal path. Numerical results are presented to show that our approach produces feasible paths that are near-optimal and satisfy the covariance bounds.