| Popis: |
This paper presents a UAV path planning strategy for optimizing the return over a finite-time horizon, where the return is specified by a bounded differentiable reward function. We represent paths using a simulated chain in a force field, where the forces are influenced by the reward function. The chain adapts continuously to changes in the return function, and produces good paths with minimal computational overhead. We compare the chain-based path planner to a look-ahead planner and extend this approach to multiple UAVs in a centralized manner. |
