ManiSkill2: A Unified Benchmark for Generalizable Manipulation Skills
Jiayuan Gu, Fanbo Xiang, Xuanlin Li, Zhan Ling, Xiqiang Liu, Tongzhou, Mu, Yihe Tang, Stone Tao, Xinyue Wei, Yunchao Yao, Xiaodi Yuan, Pengwei Xie,, Zhiao Huang, Rui Chen, Hao Su

TL;DR
ManiSkill2 is a comprehensive, dynamic simulation benchmark for generalizable manipulation skills, supporting diverse tasks, object variations, and learning algorithms, with optimized infrastructure for fast visual input learning.
Contribution
It introduces ManiSkill2, a unified, scalable benchmark with diverse manipulation tasks, object models, and evaluation protocols, advancing research in embodied AI manipulation skills.
Findings
Supports 20 manipulation task families with 2000+ objects
Enables CNN-based policies to learn at 2000 FPS on a standard workstation
Provides open-source code and an online challenge for community engagement
Abstract
Generalizable manipulation skills, which can be composed to tackle long-horizon and complex daily chores, are one of the cornerstones of Embodied AI. However, existing benchmarks, mostly composed of a suite of simulatable environments, are insufficient to push cutting-edge research works because they lack object-level topological and geometric variations, are not based on fully dynamic simulation, or are short of native support for multiple types of manipulation tasks. To this end, we present ManiSkill2, the next generation of the SAPIEN ManiSkill benchmark, to address critical pain points often encountered by researchers when using benchmarks for generalizable manipulation skills. ManiSkill2 includes 20 manipulation task families with 2000+ object models and 4M+ demonstration frames, which cover stationary/mobile-base, single/dual-arm, and rigid/soft-body manipulation tasks with…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
Taxonomy
TopicsRobot Manipulation and Learning · Reinforcement Learning in Robotics · Human Pose and Action Recognition
