Research map / Multi-agent systems and games / Human-AI and goal-conditioned
MoCapAct: A Multi-Task Dataset for Simulated Humanoid Control
Nolan Wagener, Andrey Kolobov, Felipe Vieira Frujeri, Ricky Loynd, Ching-An Cheng, Matthew Hausknecht
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add reward-weighted regression instead of unweighted behavioral cloning (ablation) | Multi-clip tracking policy · CMU MoCap snippets | average normalized episode reward: 0.654 ± 0.005 -> 0.688 ± 0.002; 5% improvement over BC | Improved |
| Add clip-weighted regression instead of unweighted behavioral cloning (ablation) | Multi-clip tracking policy · CMU MoCap snippets | average normalized episode reward: 0.654 ± 0.005 -> 0.671 ± 0.003; 3% improvement over BC | Improved |
| Add advantage-weighted regression instead of unweighted behavioral cloning (ablation) | Multi-clip tracking policy · CMU MoCap snippets | average normalized episode reward: 0.654 ± 0.005 -> 0.661 ± 0.003; 1% improvement over BC | Improved |
| Reuse a pretrained general low-level policy rather than train with no low-level policy (ablation) | Humanoid reinforcement-learning transfer · Go-to-target | return: 7.5 ± 1.1 -> 96.3 ± 2.8 | Improved |
| Restrict the pretrained low-level policy to locomotion clips rather than use the general policy (ablation) | Humanoid reinforcement-learning transfer · Go-to-target | return: 96.3 ± 2.8 -> 66.1 ± 32.8 | Hurt |
| Reuse a pretrained general low-level policy rather than train with no low-level policy (ablation) | Humanoid reinforcement-learning transfer · Velocity control | return: 1157 ± 89 vs. 1074 ± 55 | Mixed |
| Restrict the pretrained low-level policy to locomotion clips rather than use the general policy (ablation) | Humanoid reinforcement-learning transfer · Velocity control | return: 1074 ± 55 -> 884 ± 81 | Hurt |
| Use stochastic expert rollouts with action noise rather than deterministic expert behavior (ablation) | Clip snippet experts · CMU MoCap snippets | average normalized episode reward: deterministic mean 0.816; noisy mean 0.689 | Hurt |
| Increase the intention prior autoregressive parameter from α = 0 to α = 0.95 (ablation) | RWR multi-clip policy · CMU MoCap snippets | policy performance: 3% improvement | Improved |
Closest papers by meaning
- Hierarchical World Models as Visual Whole-Body Humanoid Controllers (ICLR 2025 · similarity 0.88)
- H-GAP: Humanoid Control with a Generalist Planner (ICLR 2024 · similarity 0.88)
- Universal Humanoid Motion Representations for Physics-Based Control (ICLR 2024 · similarity 0.88)
- Humanoid Horizon: Extending Task Horizon in Whole-Body Loco-Manipulation via Parallel Training, Dynamic Starting, and Reward Gating (NeurIPS 2026 · similarity 0.87)
- Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning (NeurIPS 2025 · similarity 0.87)
- MotionPRO: Exploring the Role of Pressure in Human MoCap and Beyond (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
