The LiReN model uses reinforcement learning and is capable of general goal-conditioned visual navigation. It is similar to existing General Navigation Models, but it leverages the conservative ...