Real Robot Challenge 2022: Learning Dexterous Manipulation from Offline Data in the Real World

TL;DR

The Real Robot Challenge 2022 facilitated remote experimentation with real robots for offline reinforcement learning, enabling the development and benchmarking of dexterous manipulation algorithms directly on physical hardware.

Contribution

This paper introduces a novel competition framework that bridges RL and robotics, providing datasets, simulation tools, and evaluation on real hardware for dexterous manipulation tasks.

Findings

Winning methods effectively learned from offline datasets.

Benchmark algorithms underperformed compared to top solutions.

The competition demonstrated the feasibility of offline RL on real robots.

Abstract

Experimentation on real robots is demanding in terms of time and costs. For this reason, a large part of the reinforcement learning (RL) community uses simulators to develop and benchmark algorithms. However, insights gained in simulation do not necessarily translate to real robots, in particular for tasks involving complex interactions with the environment. The Real Robot Challenge 2022 therefore served as a bridge between the RL and robotics communities by allowing participants to experiment remotely with a real robot - as easily as in simulation. In the last years, offline reinforcement learning has matured into a promising paradigm for learning from pre-collected datasets, alleviating the reliance on expensive online interactions. We therefore asked the participants to learn two dexterous manipulation tasks involving pushing, grasping, and in-hand orientation from provided real-robot datasets. An extensive software documentation and an initial stage based on a simulation of the real set-up made the competition particularly accessible. By giving each team plenty of access budget to evaluate their offline-learned policies on a cluster of seven identical real TriFinger platforms, we organized an exciting competition for machine learners and roboticists alike. In this work we state the rules of the competition, present the methods used by the winning teams and compare their results with a benchmark of state-of-the-art offline RL algorithms on the challenge datasets.