Value-Based Reinforcement Learning for Continuous Control Robotic Manipulation in Multi-Task Sparse Reward Settings

TL;DR

This paper demonstrates that value-based reinforcement learning, specifically RBF-DQN, can effectively and efficiently learn continuous robotic manipulation tasks in multi-task sparse reward environments, outperforming some policy-gradient methods.

Contribution

The paper introduces RBF-DQN for continuous control in sparse reward settings and shows its superior convergence speed over existing algorithms like TD3, SAC, and PPO.

Findings

RBF-DQN converges faster than TD3, SAC, and PPO on robotic tasks.

Hindsight Experience Replay and Prioritized Experience Replay enhance RBF-DQN performance.

Value-based methods are more sensitive to data augmentation and replay techniques than policy-gradient methods.

Abstract

Learning continuous control in high-dimensional sparse reward settings, such as robotic manipulation, is a challenging problem due to the number of samples often required to obtain accurate optimal value and policy estimates. While many deep reinforcement learning methods have aimed at improving sample efficiency through replay or improved exploration techniques, state of the art actor-critic and policy gradient methods still suffer from the hard exploration problem in sparse reward settings. Motivated by recent successes of value-based methods for approximating state-action values, like RBF-DQN, we explore the potential of value-based reinforcement learning for learning continuous robotic manipulation tasks in multi-task sparse reward settings. On robotic manipulation tasks, we empirically show RBF-DQN converges faster than current state of the art algorithms such as TD3, SAC, and PPO. We also perform ablation studies with RBF-DQN and have shown that some enhancement techniques for vanilla Deep Q learning such as Hindsight Experience Replay (HER) and Prioritized Experience Replay (PER) can also be applied to RBF-DQN. Our experimental analysis suggests that value-based approaches may be more sensitive to data augmentation and replay buffer sample techniques than policy-gradient methods, and that the benefits of these methods for robot manipulation are heavily dependent on the transition dynamics of generated subgoal states.