KlaskTron-IL: A Contact-Rich Adversarial Imitation Learning Benchmark with a Validated Sim-to-Real Bridge
cs.RO
Submitted: 2026-11-02
Updated: 2026-11-02
Code: https://github.com/IDSCETHZurich/klasktron-il
Project page: https://www.klaskgame.com/pages/about-us
License: http://creativecommons.org/licenses/by/4.0/
The gist: KlaskTron is an open-hardware physical platform for two-player adversarial reinforcement learning, with a learned actuator model that supports zero-shot sim-to-real transfer of trained policies.
Terminology
Abstract
KlaskTron is an open-hardware physical platform for two-player adversarial reinforcement learning, with a learned actuator model that supports zero-shot sim-to-real transfer of trained policies. Building on this platform, we contribute KlaskTron-IL: a multi-tier imitation learning dataset (80k human demonstration timesteps collected directly on the physical robot, 320k symmetry-augmented timesteps, and synthetic expert trajectories up to 5M steps generated by the zero-shot-validated PPO policy), and diagnostic baselines that expose three characteristic failure modes of standard IL methods under contact-rich adversarial dynamics: BC's data scaling plateau, GAIL's interpretable but weak-gradient discriminator (Spearman ρ= -1.0 peg-ball proximity recovery, but a logit span under 0.1), and an asymmetric BC-prior x reward-density interaction. Because the policy interface used throughout this paper is the same one the actuator model bridges to hardware, a policy trained on this benchmark can run unchanged on the physical robot, as we demonstrate zero-shot for the strongest BC baseline, so long as its commands respect the actuator model's plus or minus 0.2 m/s identification range.
Sources
- D4RL: Datasets for Deep Data-Driven Reinforcement Learning
- AiRLIHockey: Highly Reactive Contact Control and Stochastic Optimal Shooting
- Dota 2 with Large Scale Deep Reinforcement Learning
Related papers
- Rethinking Actions: Continuity-Aware Priors for World-Model Planning
- MOSAIC: Modular Obstacle and Spatial-Aware Framework for Manipulation In Cluttered Scenes
- SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning
- DexSeed: Expanding Human Hand-Object Interactions into Robot Demonstrations for Dexterous Policy Learning
- LaST0.5: A Generalist World Action Model with Interaction-Centric Latent Reasoning
- CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation