Human-like path planning in the presence of landmarks

This work proposes a path planning algorithm for scenarios where the agent has to move strictly inside the space defined by signal emitting bases. Considering a base can emit within a limited area, it is necessary for the agent to be in the vicinity of at least one base at each point along the path in order to receive a signal. The algorithm starts with forming a specific network, based on the starting point such that only the bases which allow the described motion are included. A second step is based on RRT*, where each edge is created solving an optimal control problem that at the end provides a human-like path. Finally the best path is selected among all the ones that reach the goal region with the minimum cost.