Efficient Focus Autoencoders for Fast Autonomous Flight in Intricate Wild Scenarios

Abstract

The autonomous navigation of aerial robots in unknown and complex outdoor environments is a challenging problem that typically requires planners to generate collision-free trajectories based on human expert rules for fast navigation. Presently, aerial robots suffer from high latency in acquiring environmental information, which limits the control strategies that the vehicle can implement. In this study, we proposed the SAC_FAE algorithm for high-speed navigation in complex environments using deep reinforcement learning (DRL) policies. Our approach consisted of a soft actor–critic (SAC) algorithm and a focus autoencoder (FAE). Our end-to-end DRL navigation policy enabled a flying robot to efficiently accomplish navigation tasks without prior map information by relying solely on the front-end depth frames and its own pose information. The proposed algorithm outperformed existing trajectory-based optimization approaches at flight speeds exceeding 3 m/s in multiple testing environments, which demonstrates its effectiveness and efficiency.

A survey on autonomous environmental monitoring approaches: towards unifying active sensing and reinforcement learning

“…The sparse reward problem refers to systems in which an agent may only receive rewards in certain scenarios. Like in collision avoidance, for example, if the agent does not meet obstacles often then it may not learn how to avoid them Hu et al (2023) . Sparse rewards might also be inherent problems of the operation environment themselves, such as in an underwater environment.…”