# It Takes Two: Learning to Plan for Human-Robot Cooperative Carrying

arXiv: https://arxiv.org/abs/2209.12890 · HTML: https://mlsanity.com/paper/2209.12890/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 24 references that are in the map.
- First posted on arXiv 26 Sep 2022.

## Builds on (most cited first, 13 of 13)

- [Learning Latent Dynamics for Planning from Pixels](https://mlsanity.com/paper/1811.04551/) (2018)
- [AI Choreographer: Music Conditioned 3D Dance Generation with AIST++](https://mlsanity.com/paper/2101.08779/) (2021)
- [A Recurrent Latent Variable Model for Sequential Data](https://mlsanity.com/paper/1506.02216/) (2015)
- [On the Utility of Learning about Humans for Human-AI Coordination](https://mlsanity.com/paper/1910.05789/) (2019)
- [Open Problems in Cooperative AI](https://mlsanity.com/paper/2012.08630/) (2020)
- [Learning to Listen: Modeling Non-Deterministic Dyadic Facial Motion](https://mlsanity.com/paper/2204.08451/) (2022)
- [Multimodal Probabilistic Model-Based Planning for Human-Robot Interaction](https://mlsanity.com/paper/1710.09483/) (2017)
- [Hierarchical Game-Theoretic Planning for Autonomous Vehicles](https://mlsanity.com/paper/1810.05766/) (2018)
- [Robot Motion Planning in Learned Latent Spaces](https://mlsanity.com/paper/1807.10366/) (2018)
- [An Evaluation of Trajectory Prediction Approaches and Notes on the TrajNet Benchmark](https://mlsanity.com/paper/1805.07663/) (2018)
- [Social-VRNN: One-Shot Multi-modal Trajectory Prediction for Interacting Pedestrians](https://mlsanity.com/paper/2010.09056/) (2020)
- [A Framework for Probabilistic Generic Traffic Scene Prediction](https://mlsanity.com/paper/1810.12506/) (2018)
- [A Sensorimotor Reinforcement Learning Framework for Physical Human-Robot Interaction](https://mlsanity.com/paper/1607.07939/) (2016)

## Built on by (most cited first, 12 of 12)

- [TACO: Benchmarking Generalizable Bimanual Tool-ACtion-Object Understanding](https://mlsanity.com/paper/2401.08399/) (2024)
- [CORE4D: A 4D Human-Object-Human Interaction Dataset for Collaborative Object REarrangement](https://mlsanity.com/paper/2406.19353/) (2024)
- [Stabilize to Act: Learning to Coordinate for Bimanual Manipulation](https://mlsanity.com/paper/2309.01087/) (2023)
- [Large Language Models for Orchestrating Bimanual Robots](https://mlsanity.com/paper/2404.02018/) (2024)
- [Enhancing Generalizable 6D Pose Tracking of an In-Hand Object with Tactile Sensing](https://mlsanity.com/paper/2210.04026/) (2022)
- [Graph Neural Networks for Decentralized Multi-Agent Perimeter Defense](https://mlsanity.com/paper/2301.09689/) (2023)
- [Autonomous Human-Robot Interaction via Operator Imitation](https://mlsanity.com/paper/2504.02724/) (2025)
- [PolySLGen: Online Multimodal Speaking-Listening Reaction Generation in Polyadic Interaction](https://mlsanity.com/paper/2604.08125/) (2026)
- [REPeat: A Real2Sim2Real Approach for Pre-acquisition of Soft Food Items in Robot-assisted Feeding](https://mlsanity.com/paper/2410.10017/) (2024)
- [The Social Context of Human-Robot Interactions](https://mlsanity.com/paper/2508.13982/) (2025)
- [Moving Out: Physically-grounded Human-AI Collaboration](https://mlsanity.com/paper/2507.18623/) (2025)
- [Learning from Humans for Proactive Assistance in Human-Robot Collaborative Transport](https://mlsanity.com/paper/2609.25351/) (2026)

## Abstract (arXiv)

Cooperative table-carrying is a complex task due to the continuous nature of the action and state-spaces, multimodality of strategies, and the need for instantaneous adaptation to other agents. In this work, we present a method for predicting realistic motion plans for cooperative human-robot teams on the task. Using a Variational Recurrent Neural Network (VRNN) to model the variation in the trajectory of a human-robot team across time, we are able to capture the distribution over the team's future states while leveraging information from interaction history. The key to our approach is leveraging human demonstration data to generate trajectories that synergize well with humans during test time in a receding horizon fashion. Comparison between a baseline, sampling-based planner RRT (Rapidly-exploring Random Trees) and the VRNN planner in centralized planning shows that the VRNN generates motion more similar to the distribution of human-human demonstrations than the RRT. Results in a human-in-the-loop user study show that the VRNN planner outperforms decentralized RRT on task-related metrics, and is significantly more likely to be perceived as human than the RRT planner. Finally, we demonstrate the VRNN planner on a real robot paired with a human teleoperating another robot.