Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Deep Imitation Learning for Bimanual Robotic Manipulation

This repository is for the NeurIPS 2020 paper "Deep Imitation Learning for Bimanual Robotic Manipulation" ( We present a deep imitation learning framework for robotic bimanual manipulation in a continuous state-action space. Our model learns to imitate table lifting movements in simulation and generalize the learned skills to tables at different locations.

The repository contains the files we used to generate our training data using and the models we used to obtain our experiment results. We used Pybullet to generate our simulations and Pytorch to build our deep learning models. More details are found below.


We created the following tasks in PyBullet to generate training data. The GIFs below are one instance of our demonstration for each of the two tasks. A video file showing our model predictions can be found in ""

Table Lifting Task Demonstration

Peg-In-Hole Task Demonstration

Generating Data

All code for generating training data is found in the "Simulation Code" folder. Simulations are built up using low-level primitives such as lift or push. All simulations are written using the PyBullet physics engine. All components in the simulations are built-up using URDF files. These files can be easily extended to create other objects/tasks to be used in simulations. In order to save time, simulations are best run on a workstation with a GPU.

Prediction Models

The prediction models are titled "Table_Lift_HDR-IL" for the table lifting task. Details about training the model can be found in "Model_Readme.txt"


To cite our work:

@misc{xie2020deep, title={Deep Imitation Learning for Bimanual Robotic Manipulation}, author={Fan Xie and Alexander Chowdhury and M. Clara De Paolis Kaluza and Linfeng Zhao and Lawson L. S. Wong and Rose Yu}, year={2020}, eprint={2010.05134}, archivePrefix={arXiv}, primaryClass={cs.RO} }


Fan Xie, Alex Chowdhury


No description, website, or topics provided.






No releases published


No packages published

Contributors 4