Robotic Table Tennis with Model-Free Reinforcement Learning
This addresses robotic control for dynamic tasks like table tennis, though it is incremental as it applies existing methods (evolutionary search and curriculum learning) to a specific domain.
The paper tackles robotic table tennis ball returns using model-free reinforcement learning, achieving an 80% return rate with policies that develop forehand and backhand strokes without architectural priors.
We propose a model-free algorithm for learning efficient policies capable of returning table tennis balls by controlling robot joints at a rate of 100Hz. We demonstrate that evolutionary search (ES) methods acting on CNN-based policy architectures for non-visual inputs and convolving across time learn compact controllers leading to smooth motions. Furthermore, we show that with appropriately tuned curriculum learning on the task and rewards, policies are capable of developing multi-modal styles, specifically forehand and backhand stroke, whilst achieving 80\% return rate on a wide range of ball throws. We observe that multi-modality does not require any architectural priors, such as multi-head architectures or hierarchical policies.