DevGRU: Depth-guided Visual Navigation using a Collision-aware Recurrent Model

To address this problem, we propose a depth image- and point-goal-conditioned navigation system, DevGRU. The proposed system employs an action predictor (AP) that generates collision-aware future trajectories, enabling effective avoidance of immediate obstacles. In conjunction with a collision predictor, the AP further compensates for errors accumulated in the goal pose estimation and proactively mitigates future deviations.

Review the primary source, validate the main result, and establish whether any listed-company transmission is direct.