2020/04/27 by Adrien Ecoffet, Joost Huizinga, Joel Lehman +2 · 2 voices · 3 citations
Computer Science · Decision Sciences · #Advanced Bandit Algorithms Research #Artificial Intelligence in Games #Forgetting #Function (biology) #Reinforcement Learning in Robotics #Reinforcement learning #Robotics #Simple (philosophy) #State (computer science) #cs.AI
paper · pdf · doi:10.1038/s41586-020-03157-9
published as Nature 590, 580-586 (2021) · 47 pages, 14 figures, 4 tables; reorganized sections and modified SI text extensively; added reference to the published version, changed title to published title; added reference to published unformatted pdf
arxiv published 2020/04/27 · openalex created_date 2020/05/01 · openalex publication_date 2021/02/24 · arxiv created 2021/09/16 · arxiv updated 2021/09/17 · openalex updated_date 2026/08/05
The promise of reinforcement learning is to solve complex sequential decision problems autonomously by specifying a high-level reward function only. However, reinforcement learning algorithms struggle when, as is often the case, simple and intuitive rewards provide sparse and deceptive feedback. Avoiding these pitfalls requires thoroughly exploring the environment, but creating algorithms that can do so remains one of the central challenges of the field. We hypothesise that the main impediment to effective exploration originates from algorithms forgetting how to reach previously visited states ("detachment") and from failing to first return to a state before exploring from it ("derailment"). We introduce Go-Explore, a family of algorithms that addresses these two challenges directly through the simple principles of explicitly remembering promising states and first returning to such states before intentionally exploring. Go-Explore solves all heretofore unsolved Atari games and surpasses the state of the art on all hard-exploration games, with orders of magnitude improvements on the grand challenges Montezuma's Revenge and Pitfall. We also demonstrate the practical potential of Go-Explore on a sparse-reward pick-and-place robotics task. Additionally, we show that adding a goal-conditioned policy can further improve Go-Explore's exploration efficiency and enable it to handle stochasticity throughout training. The substantial performance gains from Go-Explore suggest that the simple principles of remembering states, returning to them, and exploring from them are a powerful and general approach to exploration, an insight that may prove critical to the creation of truly intelligent learning agents.