Research built with ARX
25 public projects ran on ARX hardware. ARX lists 21 of them on arx-x.com; in 10 the paper, project page or code names ARX hardware itself, and we link to that place.
Labs using ARX
Physical Intelligence, Stanford, Carnegie Mellon, USC, HKUST, Tsinghua, Peking University, Shanghai Jiao Tong, Zhejiang University, Beijing Institute of Technology, SUSTech.
Published something on ARX hardware? Send the link to info@arx-global.us and we'll add it.
2026
| Date | Project | ARX hardware |
|---|---|---|
| Jul 2026 | ModPack Open-source teleoperation framework for two-armed mobile robots, built around a wearable backpack. Stanford | Uses a pair of 6-DoF ARX5 leader arms |
| Jul 2026 | Xiaomi-Robotics-1 Robot foundation model pre-trained on 100K+ hours of UMI data, post-trained on 7,200+ hours of robot data. Xiaomi | Listed by ARX |
| Jul 2026 | Xiaomi-Robotics-U0 38B autoregressive world foundation model for embodied intelligence. Xiaomi | Listed by ARX |
| Jul 2026 | LingBot-VA 2.0 Video-action model with native video-action pretraining for robot control. Robbyant (Ant Group) | Listed by ARX |
| Jul 2026 | HELP Human-efficient VLA post-training: two operators supervise twelve robots using rollout segmentation. | Listed by ARX |
| Jul 2026 | EgoWAM Trains manipulation policies on human egocentric video by predicting actions and scene change. Georgia Tech | Listed by ARX |
| Apr 2026 | π0.7 Physical Intelligence's April 2026 generalist robot model. Physical Intelligence | Listed by ARX |
| Apr 2026 | SIM1 Physics-aligned cloth simulator that turns about 200 human demonstrations into synthetic training data. Shanghai AI Lab | Listed by ARX |
| Feb 2026 | Xiaomi-Robotics-0 4.7B-parameter vision-language-action model (Qwen3-VL-4B with a diffusion transformer). Xiaomi | Listed by ARX |
2025
| Date | Project | ARX hardware |
|---|---|---|
| Dec 2025 | Motus Unified latent-action world model for understanding, action and video generation. Tsinghua University | Real-world experiments on AC-One |
| Nov 2025 | π*0.6 A VLA that improves from its own experience with reinforcement learning. Physical Intelligence | Listed by ARX |
| Nov 2025 | Human to Robot Research on transferring skills from human data to robots. Physical Intelligence | Listed by ARX |
| Oct 2025 | RoboChallenge Evaluation platform that runs submitted policies on physical robots. RoboChallenge | Platforms include Arx5 |
| Sep 2025 | InternVLA Policy model that unifies vision-language understanding, visual foresight and action generation. Shanghai AI Lab | Listed by ARX |
| Jul 2025 | Vidar Embodied video diffusion model for bimanual manipulation, adapted to robots with masked inverse dynamics. | Listed by ARX |
| Jun 2025 | ViA Active perception for bimanual manipulation, learned from human demonstrations. Stanford (CoRL 2025) | Acknowledges ARX for the robot hardware |
| May 2025 | SimpleVLA-RL Reinforcement-learning framework for vision-language-action models. PRIME-RL | Listed by ARX |
| Apr 2025 | π0.5 Vision-language-action model aimed at open-world generalization. Physical Intelligence | Listed by ARX |
2024
| Date | Project | ARX hardware |
|---|---|---|
| Oct 2024 | π0 Generalist vision-language-action flow model. Physical Intelligence | Training mixture includes "Bimanual ARX" |
| Oct 2024 | RDT 1.2B-parameter diffusion foundation model for bimanual manipulation (RDT-1B). Tsinghua University | Listed by ARX |
| Jul 2024 | UMI on Legs Manipulation policies made mobile on a legged robot with whole-body control (CoRL 2024). Stanford, Columbia, Google DeepMind | Listed as a project using the ARX5 SDK |
Also using ARX arms
| Year | Project | ARX hardware |
|---|---|---|
| 2026 | EgoVerse Platform and dataset of 1,362 hours of egocentric human demonstrations. Georgia Tech, Stanford, UC San Diego, ETH Zürich and others | Both robot setups use two 6-DoF ARX5 arms |
| 2025 | RoboTwin 2.0 Simulation framework and 50-task benchmark for bimanual manipulation. SJTU, HKU MMLab, Shanghai AI Lab and others | ARX-X5 is one of the simulated embodiments |
| 2024 | UMI Universal Manipulation Interface: handheld grippers for collecting in-the-wild demonstrations. Stanford, Columbia, Toyota Research Institute | Deployment code for ARX arms (umi-arx) listed in arx5-sdk |
| 2025 | UVA Unified video-action model. Stanford | Deployment code for ARX arms listed in arx5-sdk |
Last updated October 10, 2026 · facts from the sources listed on this page
Use the same hardware.
Quotes for the arms and systems behind this work.