RoTO: Robot Tactile Olympiad
A reinforcement learning benchmark environment to standardise and promote research in tactile-based manipulation (simulated in NVIDIA Isaac Lab).
Introduced in NeurIPS 2025 Paper project page →
A reinforcement learning benchmark environment to standardise and promote research in tactile-based manipulation (simulated in NVIDIA Isaac Lab).
Introduced in NeurIPS 2025 Paper project page →
Tactile-based reinforcement learning (RL) is currently hindered by fragmented research and a focus on over-saturated orientation tasks. We introduce the Robot Tactile Olympiad (roto), a GPU-parallelised benchmark designed to standardise tactile-based RL across three distinct robotic morphologies (10-DOF to 24-DOF). Unlike prior benchmarks, \texttt{roto} focuses on end-to-end ``blind'' manipulation, utilising only proprioception and tactile sensing without state information or privileged teacher-student distillation.
We demonstrate a significant performance leap, with our blind agents achieving 13 Baoding ball rotations in 10 seconds, an order of magnitude faster than current state-of-the-art speeds. By open-sourcing our environments and robustly tuned baselines, we reduce the barrier to entry and enable researchers to prioritise fundamental algorithmic challenges over tedious RL tuning.
Visitors around the world
If you use RoTO or this work, please cite:
@inproceedings{miller2025tactilerl,
author = {Miller, Elle and McInroe, Trevor and Abel, David and Mac Aodha, Oisin and Vijayakumar, Sethu},
title = {Enhancing Tactile-based Reinforcement Learning for Robotic Control},
booktitle = {NeurIPS},
year = {2025},
}