HOI-Retarget:
Contact-Centric Retargeting
for Human-Object Interaction

A contact-centric method that transfers human–object interaction to humanoid embodiments while explicitly preserving the location and timing of object contacts.

Jihwan Shin · Adrià López Escoriza · Junzhe He · Matthias Heyrman · Marco Hutter

Robotic Systems Lab, ETH Zürich

Humanoid robots carrying, lifting and moving a variety of everyday objects, retargeted from human motion capture.
HOI-Retarget preserves object-relative contacts while generating robot interaction references across diverse motions, objects and humanoid platforms.

Abstract

Learning from demonstration (LfD) has enabled humanoid robots to acquire diverse whole-body skills, but extending this paradigm to human-object interaction (HOI) is limited by the availability of robot-compatible interaction references. We present HOI-Retarget, a contact-centric retargeting method that transfers HOI onto a humanoid robot for large-scale motion-data generation. Its windowed trajectory optimization uses every labeled contact as a target in the object frame, balancing body tracking, foot support and smoothness under the robot's kinematic limits. The method can augment a single demonstration across object sizes, absorb contacts reconstructed from monocular video, and extend to several robots manipulating one object. We publicly release the code and the retargeted motion dataset.

0.5 cm mean contact-point gap, down from 18.3 cm for the closest baseline
4.6× faster to retarget a clip than the interaction-aware baseline
6,952 released robot HOI clips — 13.8 hours over 75 unique objects

Method

How it works

Pipeline: source HOI motion, IK retargeting with object scaling, windowed trajectory optimization, and downstream policy use.
(a) A captured or video-reconstructed clip supplies human motion, an object trajectory and contact labels. (b) IK retargeting maps the human onto the robot; the object mesh and trajectory scale by the robot-to-human height ratio, carrying the contact targets with them. (c) A windowed trajectory optimization recovers those contacts under the robot's kinematic limits. (d) The result drives downstream policies.

Results

Diverse objects and embodiments

OMOMO retargets on the Unitree G1 and H2, one camera distance across all three panels.

Human sourceUnitree G1Unitree H2

All three panels share one camera distance, so the heights are directly comparable: the H2 stands about as tall as the human, the G1 well short of both.

Source dataset & robots: OMOMO (Li et al., 2023)·Unitree G1·Unitree H2

Generality

Beyond a single dataset

Four more HOI datasets, and two-person demonstrations retargeted against the shared object trajectory.

Human sourceUnitree G1Unitree H2

Three single-actor clips from public HOI datasets, plus two collaborative sequences in which a pair of robots share one object. Four datasets join OMOMO in the released corpus — ParaHome, NeuralDome, CoRoleHOI and IMHD2; the HUMOTO clip is shown as a retargeting result but is not part of the release.

Source datasets: IMHD2 (Zhao et al., 2024)·NeuralDome (Zhang et al., 2023)·HUMOTO (Lu et al., 2025)·CoRoleHOI (Zuo, 2026)·ParaHome (Kim et al., 2024)

Data generation

Object-scale augmentation

One captured motion becomes many. The contact targets live in the object frame, so resizing the object carries them with its surface and the clip re-solves at any size without re-annotating the interaction.

×0.25×0.50×1.00×1.25×1.50
G1H2

One source clip at five object scales, from a quarter to one-and-a-half times true size, on both robots. The grasp re-anchors to the new surface at every scale.

Source dataset & robots: OMOMO (Li et al., 2023)·Unitree G1·Unitree H2

In the wild

From monocular video

A table in none of the datasets above, recorded on a single camera and reconstructed by CARI4D. Noisy per-frame contacts are collapsed to one point per sticking segment and corrected before the same optimization runs.

Captured videoReconstructionHOI-RetargetAfter refinement

A hand-held capture of a person moving a table the method has never seen, reconstructed into an SMPL-X body and a 6-DoF object track, retargeted with the contact-correction tool, then refined into a dynamically feasible trajectory. The subject's face is blurred throughout.

Reconstruction: CARI4D (Xie et al., 2026)·Unitree G1

Downstream

Dynamic refinement

A kinematic retarget is not yet physically feasible. Two refiners of different families — a per-clip RL tracker and MuJoCo-based SBTO — start from OmniRetarget's references and from ours, on the same budget.

Kinematic referenceTrajectory optimizationRL tracker

The kinematic retarget is a reference, not a controller. Both refiners start from the left panel; each produces a dynamically feasible trajectory usable for reference-state initialization downstream.

Refiner & data: DynaRetarget (Dhedin et al., 2026)·OMOMO (Li et al., 2023)

Release

Dataset

Kinematic humanoid interaction references from five source datasets, retargeted to two humanoid platforms, including synchronized two-robot collaboration.

6,952robot HOI clips
13.8 hof retargeted motion
75unique objects
2humanoid platforms

BibTeX

Replace the note with the arXiv identifier once the preprint is up.

@article{shin2026hoiretarget,
  title  = {HOI-Retarget: Contact-Centric Retargeting for Human-Object Interaction},
  author = {Shin, Jihwan and L\'opez Escoriza, Adri\`a and He, Junzhe and
            Heyrman, Matthias and Hutter, Marco},
  year   = {2026},
  note   = {Manuscript under review}
}