Reward-Based Environment States for Robot Manipulation Policy Learning
This addresses the challenge of efficient policy learning for robot manipulation, though it appears incremental as it builds on existing deep reinforcement learning methods with a new state representation.
The paper tackled the problem of training robot manipulation policies by proposing a novel state representation based on rewards predicted from an image-based task success classifier, achieving up to 97% task success in simulation with a Pepper robot on a grab-and-lift task.
Training robot manipulation policies is a challenging and open problem in robotics and artificial intelligence. In this paper we propose a novel and compact state representation based on the rewards predicted from an image-based task success classifier. Our experiments, using the Pepper robot in simulation with two deep reinforcement learning algorithms on a grab-and-lift task, reveal that our proposed state representation can achieve up to 97% task success using our best policies.