Hi ROS community ,
Like many in the open-source community, we’ve been frustrated by the massive hardware premiums required to get into robotics and embodied AI research. Industrial AMRs and collaborative arm setups easily cross the $50k mark.
A startup, NVatom, wants to change that, so it co-developed Mobile OpenArm X1 alongside OpenArm, bringing that entry barrier down to under $9k while keeping it fully open-source and natively supported on ROS 2. It is a fully transparent, modular development platform engineered specifically for low-level control, simulation, and data collection.
NVatom managed to scale the hardware cost down significantly. For context, the base Education Edition features an industrial-graded LiDAR-guided autonomous mobile robot paired with a 16-DoF arm/gripper setup, hitting a hardware cost of $9,000
Core Specs & Tech Stack:
-
Mobility & Kinematics: 4WD omnidirectional AMR base supporting 360° spatial turning and continuous 360° waist rotation.
-
Sensing: Integrated LiDAR tracking and odometry for global localization, centimeter-level positioning, and dynamic obstacle avoidance.
-
AI / Model Training: Native spatial-action data fusion (LiDAR point clouds + joint states) optimized for training Vision-Language-Action (VLA) models.
-
Software Ecosystem: Out-of-the-box support for Hugging Face LeRobot, ACT, and Diffusion Policy, alongside simulation integration for Isaac Gym and MuJoCo.
-
Transparency: Complete access to low-level driver source code and unified APIs.
The goal in NVatom is to build an open foundation so developers can iterate faster without proprietary walls. The platform is currently up for pre-order, and the entire stack is decoupled and modular.
Love to hear your thoughts on the hardware layout. Are there specific sensor payload configurations or simulation environments you’d like to see natively supported out of the box?
Visit nvatom at www.nvatom.com or x1@nvatom.com
