Rapidly exploring Random Orienteering

Data gathering in the physical world is tedious and often risky. Robots are ideally suited for such applications. The objective of a data gathering robot is to travel between two points maximizing the information it gains while not exceeding energy or cost constraints. This problem can be formalized as that of finding budgeted routes in a graph such that the reward collected by the route is maximized, where the reward at nodes can be related. This problem is known as correlated orienteering. Unfortunately, state of the art solutions are too slow to even present an approximate solution while running online. We describe a sampling based anytime orienteering algorithm that provides approximate solutions to the orienteering problem with computational costs that enable it to run online. The key idea is to pose orienteering as a combination of a Constraint Satisfaction Problem (CSP) and a Traveling Salesman Problem (TSP). This formulation allows us to restrict the search space to routes that incur minimum distance to visit a set of selected nodes, and rapidly search this space using random sampling. We demonstrate that the state of the art methods take 10-12 minutes to solve a 400 node problem, whereas the method presented here takes 7-8 seconds.