Heave-Ho

Decentralized Long-Horizon Multi-Humanoid Collaboration
Ting-Hsuan Chen1* Siheng Zhao1*‡ Han Jiang1* Qi Pan1 Louison Lu1 Jiaheng Luo1 Henghui Bao1 Yiyu Chen2 Quan Nguyen1 Yue Wang1†
1University of Southern California2VinMotion*Equal contribution†Corresponding author‡Project lead
2identical humanoids in training
10robots at deployment
12tasks in HH-Bench
0network messages between robots

Abstract

Collaboration promises to extend humanoid capabilities beyond what individual robots can achieve. However, realizing this potential requires coordinated whole-body interactions, coherent long-horizon planning, and precise synchronization across robots. We introduce Heave-Ho, a decentralized framework for long-horizon collaborative multi-humanoid loco-manipulation. At the low level, we develop a multi-agent post-training framework that adapts a humanoid foundation tracking model to physically coupled interactions between humanoids, enabling coordinated decentralized control from local observations alone. At the high level, we train a vision-language planner and design a speech-based leader-follower coordination mechanism to synchronize decentralized execution without network communication. Trained with two identical humanoids, our system scales to teams of up to ten robots and generalizes to heterogeneous teams without retraining. To benchmark these capabilities, we introduce HH-Bench, comprising 12 tasks involving teams of 2 to 10 humanoids, including heterogeneous teams. Systematic evaluations show that Heave-Ho consistently outperforms baseline methods, while real-world experiments demonstrate robust decentralized collaboration across challenging tasks.

How it works

The leader chooses the next team skill and speaks it. Followers hear it on board, and every robot starts on a shared schedule.

Method overview of Heave-Ho

One voice, one team

Press play with the sound on. The leader speaks each command, in English and then in Mandarin. Followers hear it with on-board microphones, and every robot starts together. Either a robot or a person can lead. Spoken commands play in real time; execution is sped up and muted. The person's voice is pitch-shifted for anonymity.

A robot leads
A person leads

Real world: two humanoids

Clips marked ⏩ Sped up are played faster than real time. Faces are obscured for anonymity.

Box handover from a truck
Precise placement
L-shaped corridor
Narrow door
Narrow door, another box
Narrow door, inside view

Real world: larger and mixed teams

The G1 policy is trained with two identical robots and deployed without retraining. The PM01 runs its own policy trained with the same recipe.

Unitree G1EngineAI PM01Human
Table carry, 4 G1
Table carry, 3 G1 + human
Table carry, PM01 + human
Table carry, G1 + PM01 + human
Narrow door, G1 + human
Stair carry, G1 + human

HH-Bench in simulation

L-shaped corridor
Narrow door
Table carry, 4 G1
Table carry, 10 G1
L-shaped sofa
Stair carry
Precise placement
Box handover
Palanquin, 4 G1
Palanquin, 2 G1
Rebar
Fire hose
Box push
Millstone
Long-horizon composition

Real-world success

No network message is exchanged between robots.

Unitree G1EngineAI PM01Human
TaskTeamSuccess
Narrow door9/10
L-shaped corridor10/10
Box handover10/10
Precise placement9/10
Narrow door10/10
Stair carry7/10
Table carry8/10
Table carry9/10
Table carry10/10
Table carry10/10

High-level planning and low-level post-training

Hover over or tap the charts for details.

High-level planning

Success of the full system with different decision makers, with 10 trials per task. One decision of our planner takes 52.5 ms on a single RTX 4090, about 120 times faster than GPT-6 (Astra) through its API (6.46 s on average).

Low-level post-training

Simulation results on each task of HH-Bench, with three seeds of 512 teams per task; the last group is the average over the 12 tasks. Averaged over the 12 tasks, Heave-Ho succeeds in 94.41% of the episodes, while no baseline or ablation exceeds 28.56%. Click a method to hide or show it.

Citation

@article{chen2026heaveho,
  title   = {Heave-Ho: Decentralized Long-Horizon Multi-Humanoid Collaboration},
  author  = {Chen, Ting-Hsuan and Zhao, Siheng and Jiang, Han and Pan, Qi and Lu, Louison and Luo, Jiaheng and Bao, Henghui and Chen, Yiyu and Nguyen, Quan and Wang, Yue},
  journal = {arXiv preprint arXiv:2611.22275},
  year    = {2026}
}