There is widespread interest in the deployment of fleets of marine robots with the potential to roam the oceans freely and collect data at an unprecedented scale. This calls for the development of efficient algorithms for multiple vehicle motion planning that can take directly into account the capabilities of each vehicle as well as the environmental conditions and lend themselves to seamless integration with control and navigation systems. The latter connection is for the most part eschewed in the literature, in spite of the obvious fact that in order for the vehicles to execute the planned motions they must at a later stage navigate with great accuracy and follow the trajectories using control algorithms that take explicitly into account the dynamical constraints of the vehicles involved. Among the methods available for underwater vehicle navigation, terrain-based techniques have recently come to the fore. These techniques avoid the use of overly expensive inertial-like motion sensor units and hold considerable promise for the development of a new breed of affordable long range navigation systems. Motivated by these considerations, we tackle in the present paper the problem of multiple vehicle motion planning by taking explicitly into consideration inter-vehicle collision avoidance, together with a number of criteria that may include simultaneous times of arrival at assigned target points, energy minimization, acoustic communication constraints, and the maximization of terrain information along the vehicle paths (as measured by some appropriate criterion) for terrain-based navigation purposes.
展开▼
