[NeurIPS 2024] CLOVER: Closed-Loop Visuomotor Control with Generative Expectation for Robotic Manipulation

Movement Primitive Diffusion (MPD) is a diffusion-based imitation learning method for high-quality robotic motion generation that focuses on gentle manipulation of deformable objects.

A training and deployment pipeline for visuomotor RL policy used to enhance the locomotion of Unitree Go1 robot dog.

[CVPR2026] GeoPredict: Leveraging Predictive Kinematics and 3D Gaussian Geometry for Precise VLA Manipulation

OBEYED-VLA: Clutter-Resistant Vision-Language-Action Models through Object-Centric and Geometry Grounding

Official Tensorflow implementation of 'Goal-Conditioned End-to-End Visuomotor Control for Versatile Skill Primitives'

We use visual data alone to learn a control policy for a robotic arm by observing expert video demonstrations. We implement and test several models, accomplishing an 85% success rate for a pick-and-place task.

This repository contains the implementation, data-generation pipelines, and trained behavior-cloning models for a kitchen pick-and-place task using IsaacLab. Task: a Franka Emika manipulator picks a tomato soup can (container) from inside a fridge and places it inside the microwave cavity.

N++2 is a multi-threaded neural network simulator with network sharding and BLAS support. It was developed from 2007 to 2010 in order to realize training and realtime inference of deep autoencoder networks. With its help I created the first visual DeepRL control policies.

This repository contains the code and data used for the research titled "A lateralized motor network in order to understand adaptation to visuomotor rotation" published in Journal of Neural Engineering.