arXiv:2610.07594 · v3

BiGym 2.0: Benchmarking Learned and Agent-Developed Policies for Humanoid Household Manipulation

首次提交 2026-10-06 09:33

版本更新 2026-11-12 03:35

雷达观测 2026-11-13 14:16

摘要

Humanoid household manipulation requires the arms to act while the body balances, steps and changes posture. We present BiGym 2.0, an adaptation of BiGym for the Unitree G1 across 20 household tasks using a unified whole-body controller for demonstration and evaluation. The suite provides 60 native human virtual-reality demonstrations per task with synchronised multi-camera views and full-body execution records. We benchmark vision-language-action fine-tuning, imitation learning, demo-driven reinforcement learning, and cold-start coding agents given the interaction budget of online reinforcement learning. With the same onboard views, proprioception and whole-body controller for every method, vision-language-action fine-tuning has the highest nine-task mean, and agent-developed programs outperform every demo-driven reinforcement learning baseline on this mean and lead on bimanual reaching. Cross-workspace stacking remains open, $π_{0.5}$ stays low on pick-box, and multi-object transport is hard for imitation learning, demo-driven reinforcement learning and coding agents. All environments, human demonstrations, and evaluation traces are open-sourced at https://github.com/swirl-uk/BiGym2.

作者备注：23 pages. v3: Fig. 2 shows the torso-pitch body command used on 14 tasks; gripper named Dex1-1 throughout; seed notation