Learning Object-conditioned Exploration using Distributed Soft Actor Critic

Object navigation is defined as navigating to an object of a given label in a\ncomplex, unexplored environment. In its general form, this problem poses\nseveral challenges for Robotics: semantic exploration of unknown environments\nin search of an object and low-level control. In this work we study\nobject-guided exploration and low-level control, and present an end-to-end\ntrained navigation policy achieving a success rate of 0.68 and SPL of 0.58 on\nunseen, visually complex scans of real homes. We propose a highly scalable\nimplementation of an off-policy Reinforcement Learning algorithm, distributed\nSoft Actor Critic, which allows the system to utilize 98M experience steps in\n24 hours on 8 GPUs. Our system learns to control a differential drive mobile\nbase in simulation from a stack of high dimensional observations commonly used\non robotic platforms. The learned policy is capable of object-guided\nexploratory behaviors and low-level control learned from pure experiences in\nrealistic environments.\n