Motion Perception in Reinforcement Learning with Dynamic Objects

Abstract

In dynamic environments, learned controllers are supposed to take motion into account when selecting the action to be taken. However, in existing reinforcement learning works motion is rarely treated explicitly; it is rather assumed that the controller learns the necessary motion representation from temporal stacks of frames implicitly. In this paper, we show that for continuous control tasks learning an explicit representation of motion improves the quality of the learned controller in dynamic scenarios. We demonstrate this on common benchmark tasks (Walker, Swimmer, Hopper), on target reaching and ball catching tasks with simulated robotic arms, and on a dynamic single ball juggling task. Moreover, we find that when equipped with an appropriate network architecture, the agent can, on some tasks, learn motion features also with pure reinforcement learning, without additional supervision. Further we find that using an image difference between the current and the previous frame as an additional input leads to better results than a temporal stack of frames.

References in corpus (8)

• DeepMind Control Suite
• Learning to Navigate in Complex Environments
• Hindsight Experience Replay
• Reinforcement Learning with Unsupervised Auxiliary Tasks
• Driving Policy Transfer via Modularity and Abstraction
• MINOS: Multimodal Indoor Simulator for Navigation in Complex Environments
• Learning to Act by Predicting the Future
• Learning Actionable Representations from Visual Observations

Cited by in corpus (3)

• Balance Between Efficient and Effective Learning: Dense2Sparse Reward Shaping for Robot Manipulation with Environment Uncertainty
• Models, Pixels, and Rewards: Evaluating Design Trade-offs in Visual Model-Based Reinforcement Learning
• Investigation of Factorized Optical Flows as Mid-Level Representations