Abstract
Modern unmanned aerial vehicles (UAVs) rely on high-frequency periodic sensor measurements in order to safely operate in cluttered environments with both static and dynamic obstacles. However, periodic sensor checking operations are time and computation consuming and they are often not needed, especially in situations where the UAV can operate without violating the safety constraint (e.g., in uncluttered free space). In this paper, we introduce a computation-aware framework that limits sensor checking and replanning operations to instances in which such operations could be necessary. To this end, we propose an approach that utilizes reachability analysis to capture the future states of a UAV operating under the effects of noise and disturbance and performs self-triggered scheduling for sensor monitoring and replanning operations while guaranteeing safety. The replanning operation is further relaxed by performing an online reachable tube shrinking. This approach is supplemented with an online speed adaptation policy based on the curvature of the planned path to minimize deviation from the desired trajectory due to complex system dynamics and controller limitations. The proposed technique is validated with both simulations and experiments focusing on a quadrotor motion planning operation in environments consisting of both static and dynamic obstacles.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have