ContactBench benchmark advances everyday robot touch and manipulation skills
ContactBench 1.0: A Benchmark for Everyday Contact-Rich Robotic Manipulation
Robotics
Summary
Robots need to handle things with touch and precision to do everyday tasks, but testing how good they are at this is hard. The authors created ContactBench 1.0, a set of 36 tasks and objects that mimic real-life handling challenges, which anyone can reproduce using 3D printing and simulation. They tested some recent robot control methods and found that simply using vision doesn’t handle touch well, and adding tactile sensors still needs improvement. This benchmark helps researchers and engineers fairly compare and improve robots’ ability to manipulate objects with contacts.
What this means in practice
- •For robotic manipulation engineers: Evaluate and improve robot manipulation algorithms using standardized, reproducible contact-rich tasks and physical or simulated objects.
- •For robotics hardware developers: Test new tactile sensors or gripper designs by integrating them with tasks from the ContactBench benchmark to measure effectiveness.
Authors
Jingyi Xiang, Yixuan Wang, Jeannette Bohg, Wenzhen Yuan
Abstract
Contact-rich manipulation is essential to everyday activities, yet evaluating robots' ability to perform these activities remains challenging because of their diverse contact behaviors and the difficulty of reproducing tasks across laboratories. Drawing on observations of everyday human manipulation, we identify eight recurring skill categories that provide a practical framework for studying these capabilities. We introduce ContactBench 1.0, an extensible benchmark that instantiates these skills through 36 reference tasks and more than 40 object assets. The assets capture task-relevant geometry and contact behavior, support reproducible fabrication using consumer-grade 3D printers, and are paired with simulation models with tunable resistance loads. Controlled task variations enable evaluation of policy generalization. We evaluate recent vision-based and vision-tactile policy learning methods on six benchmark tasks. The results reveal that the evaluated vision-only policies struggle to perceive or react to contact events, while the evaluated visual-tactile policies do not meaningfully improve performance. By providing shared tasks and physical and simulated assets, ContactBench 1.0 establishes a foundation for reproducible evaluation of contact-rich manipulation research. The benchmark is open-sourced at https://jingyi-xiang.github.io/contactbench