Benchmarking Deep Reinforcement Learning for Continuous Control

TL;DR

This paper introduces a comprehensive benchmark suite for continuous control tasks in deep reinforcement learning, enabling standardized evaluation and comparison of algorithms across diverse challenging scenarios.

Contribution

It provides a new benchmark suite with diverse tasks and reference implementations, addressing the lack of standardized evaluation in continuous control RL research.

Findings

Systematic evaluation of multiple RL algorithms on the benchmark.

Identification of strengths and weaknesses of different algorithms.

Facilitation of reproducibility and future research in continuous control.

Abstract

Recently, researchers have made significant progress combining the advances in deep learning for learning feature representations with reinforcement learning. Some notable examples include training agents to play Atari games based on raw pixel data and to acquire advanced manipulation skills using raw sensory inputs. However, it has been difficult to quantify progress in the domain of continuous control due to the lack of a commonly adopted benchmark. In this work, we present a benchmark suite of continuous control tasks, including classic tasks like cart-pole swing-up, tasks with very high state and action dimensionality such as 3D humanoid locomotion, tasks with partial observations, and tasks with hierarchical structure. We report novel findings based on the systematic evaluation of a range of implemented reinforcement learning algorithms. Both the benchmark and reference implementations are released at https://github.com/rllab/rllab in order to facilitate experimental reproducibility and to encourage adoption by other researchers.