# From Digital Human Interactions to Physics-Based Humanoid Skills: Physics-Grounded Post-Training of Interaction Generators

arXiv: https://arxiv.org/abs/2610.10322 · HTML: https://mlsanity.com/paper/2610.10322/

## Key facts

- No citing papers in the map yet (data through 7 Oct 2026).
- Builds on 11 of its 46 references that are in the map.
- First posted on arXiv 7 Oct 2026.

## Builds on (most cited first, 11 of 11)

- [Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning](https://mlsanity.com/paper/2108.10470/) (2021)
- [PhysHOI: Physics-Based Imitation of Dynamic Human-Object Interaction](https://mlsanity.com/paper/2312.04393/) (2023)
- [Retargeting Matters: General Motion Retargeting for Humanoid Motion Tracking](https://mlsanity.com/paper/2510.02252/) (2025)
- [InterMask: 3D Human Interaction Generation via Collaborative Masked Modeling](https://mlsanity.com/paper/2410.10010/) (2024)
- [RL from Physical Feedback: Aligning Large Motion Models with Humanoid Control](https://mlsanity.com/paper/2506.12769/) (2025)
- [Learning Whole-Body Human-Humanoid Interaction from Human-Human Demonstrations](https://mlsanity.com/paper/2601.09518/) (2026)
- [PhysiInter: Integrating Physical Mapping for High-Fidelity Human Interaction Generation](https://mlsanity.com/paper/2506.07456/) (2025)
- [HINT: Hierarchical Interaction Modeling for Autoregressive Multi-Human Motion Generation](https://mlsanity.com/paper/2601.20383/) (2026)
- [TeamHOI: Learning a Unified Policy for Cooperative Human-Object Interactions with Any Team Size](https://mlsanity.com/paper/2603.07988/) (2026)
- [Learning to Assist: Physics-Grounded Human-Human Control via Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2603.11346/) (2026)
- [PhysMoDPO: Physically-Plausible Humanoid Motion with Preference Optimization](https://mlsanity.com/paper/2603.13228/) (2026)

## Abstract (arXiv)

Recent methods have made promising progress in generating interactions between two humanoids, largely relying on physics-based tracking policies to convert digital reference motions into executable trajectories. However, limited tracking capabilities restrict the range of reference motions that can be successfully executed, reducing data utilization. Moreover, even successful tracking does not guarantee physically plausible responses or faithful realization of the intended interactions. In this paper, we introduce DIGHT, a co-adaptive framework that couples a Digital human Interaction Generator with a Humanoid Tracking policy. Our DIGHT first executes multiple text-conditioned interaction candidates in simulation using a fixed tracker. It then constructs physics-grounded preferences from the resulting rollouts, covering both general executability and interaction fidelity. Rather than collapsing these signals into a single scalar reward for candidate ranking, we align the pretrained generator using physics-decoupled diffusion direct preference optimization (DPO), preserving criterion-specific supervision without differentiating through the simulator. To improve executability, preference pairs are derived from tracking error, friction, and floating. Additionally, to improve interaction fidelity, we propose to incorporate force feedback from simulator as a measure of contact fidelity and construct preferences over contact occurrence, location, duration, and force magnitude. The aligned generator then supplies reference motions for fine-tuning the tracker, improving compatibility between generation and physical execution. Extensive experiments demonstrate that our approach not only improves the physical plausibility of generated motions but also enables more reliable and faithful humanoid interactions in simulation.