Open AI models for robots, with the license of their code and of their weights, whether commercial use is allowed, and the robots on Robotheca that can run them. Every entry links to its source.
Page 1 of 2 · newest commit first
Training and evaluation code for Asimov 1 humanoid locomotion policies, built as an Isaac Lab extension using PPO and adversarial motion priors, for sim-to-real deployment.
An open vision-language-action model that turns camera images and language instructions into robot actions for manipulation and humanoid skills, post-trainable for new robots.
A hierarchical robot foundation model for long-horizon manipulation, pairing a world-model-guided high-level planner with a generalist low-level policy for robot embodiments.
Motus is a unified latent action world model that combines video generation, understanding and action experts to predict robot behavior, with checkpoints for RoboTwin 2.0 tasks.
A spatially guided vision-language-action model for generalist robot control, pairing a language head and an action head trained together.
LingBot-VLA 2.0 is a Vision-Language-Action foundation model for real-world robot manipulation, pre-trained on about 60,000 hours of robot and egocentric human data.
VLAC is a pair-wise critic and manipulation model for real-world robot reinforcement learning and data refinement, scoring task progress and completion from images and a task description.
A vision-language-action robot foundation model pre-trained on over 100K hours of embodiment-free UMI trajectories, then post-trained for mobile manipulation in unseen environments.
A vision-language-action foundation model for generalist robot manipulation, trained on about 37,300 hours of heterogeneous embodied data and aimed at home and industrial tasks.
MiniCPM-Robot is a family of on-device embodied AI models for robots, with a manipulation model and a target-tracking model that run locally on robot hardware.
An open vision-language-action model from Tencent that learns robot manipulation from large-scale egocentric demonstration data and transfers across robot platforms.
TD-MPC2 is an official research implementation of a scalable model-based reinforcement learning algorithm for continuous control, trained and evaluated on single or multiple tasks.
An open-source toolkit for training, running and adapting robot world models, used to evaluate robot policies by generating video rollouts.
LingBot-VA is an autoregressive video-action world model for robot control that predicts visual dynamics and actions in one interleaved sequence.
VITRA is a pretraining approach for vision-language-action robot manipulation models, learned from large-scale real-life human hand activity videos treated as dexterous end-effector data.
DreamZero is a research project from NVIDIA's GEAR Lab: a World Action Model that jointly predicts robot actions and video, aimed at zero-shot performance on unseen tasks.
RoboVLMs is a codebase for turning pretrained vision-language models into vision-language-action policies that control robots. It supports training and evaluation on CALVIN and SimplerEnv.
A PyTorch release of JEPA-WMs, joint-embedding predictive world models for physical planning, with pretrained models, datasets and training and evaluation code from Meta AI Research.
A vision-language-action model that adds a lightweight adapter bridge to a small Qwen2.5-0.5B backbone so it can learn robot control; the code is fine-tuned and evaluated on LIBERO and CALVIN.
UnifoLM-WMA-0 is Unitree's open-source world-model and action framework for general-purpose robot learning across several robot types. It can simulate robot interactions and enhance policy decisions.
RDT2 is a family of open robot foundation models that map camera images and a language instruction to robot arm actions, aiming at zero-shot use on unseen robot setups.
X-VLA is a vision-language-action model that uses soft prompts to train one Transformer backbone across many robot embodiments, with a server-client setup for inference.
A vision-language model that locates the objects and free spaces a natural-language prompt refers to, with step-by-step spatial reasoning, for robotics use.
Pelican-VL 1.0 is an open-source family of embodied multimodal brain models, from 7B to 72B parameters, that take visual and text input for robotics-oriented reasoning.
AI models 1–24 of 29