GENISOM AI debuts deployable robotics platforms at ICRA 2026
At ICRA 2026, GENISOM AI may have been new to many international attendees — but it is not a concept-stage robotics startup.

At ICRA 2026, GENISOM AI may have been new to many international attendees — but it is not a concept-stage robotics startup. Founded in December 2023, the company said it has already produced and delivered more than 10,000 units, making it one of the few robotics companies to reach production scale in under three years.
GENISOM AI said that milestone puts it in a different position than many early-stage physical AI companies entering the global market. Like Unitree, GENISOM AI is building around manufacturable platforms designed to move beyond laboratory demonstrations and reach customers at scale.
However, the Beijing-based company said it puts greater emphasis on industry deployment, combining robotics hardware and software, in-house core technologies, and real-world application capabilities.
M1, the company’s industrial-grade quadruped robot
At ICRA in Vienna last week, GENISOM AI showcased its M1 and L1 -series robots as mature, mass-produced systems.
The GENISOM M1 is the company’s industrial-grade quadruped robot, rated for a 30 kg (66.1 lb.) continuous walking payload and a payload-to-weight ratio approaching 1:1. The platform also carries an IP67 protection rating and can support up to five hours of runtime, depending on payload and operating conditions.
That payload capacity is supported in part by GENISOM AI’s in-house P85MAX-S joint actuator module, which delivers up to 180 N·m peak torque in an 86 mm (3.3 in.) diameter form factor weighing approximately 1 kg (2.2. lb.). The company said that developing actuators in-house gives it direct control over a key hardware-software integration point that shapes locomotion performance, payload capacity, and system reliability.
The GENISOM M1 Ultra adds an advanced perception layer to the series. It draws on bird’s eye view (BEV)-based temporal fusion and occupancy-network methods — approaches more commonly associated with autonomous driving perception. This approach supports the Omni-Panorama system for 720° 3D spatial awareness and a more semantic understanding of surrounding structures, obstacles, and spatial relationships.
Beyond the show floor, GENISOM AI said its robot platforms have already moved into real-world applications, including research and education, security patrols, and emergency response, with different platforms serving different needs.
On the research and development side, the GENISOM L1 EDU combines NVIDIA Jetson Orin NX computing with Livox Mid-360 lidar, RealSense depth cameras, GNSS, and 5G connectivity. It is designed to support simultaneous localization and mapping (SLAM), 3D reconstruction, autonomous navigation research, and open-source development workflows.
Source: The Robot Report