Beijing, China, June 10, 2026 (GLOBE NEWSWIRE) — GENISOM AI, a full-stack technology provider in embodied intelligence, made its debut at ICRA 2026 in Vienna. The company presented its complete embodied intelligence ecosystem, anchored by the GENISOM M1 quadruped robot, alongside in-house joint actuator modules, simulation infrastructure, autonomous navigation systems, and an AI agent framework designed for real-world deployment.
GENISOM AI Made Its First Appearance at ICRA
ICRA is widely recognized as one of the world’s leading conferences on robotics and automation, drawing researchers, engineers, and industry partners from across the globe.
For GENISOM AI, Vienna marked the company’s first major exhibition at the conference, and it arrived not with a single announcement, but with a full hardware and software lineup representing the depth of its embodied intelligence stack.
At ICRA 2026, GENISOM AI showcased its GENISOM M1, L1, L1-W, and L1 EDU robot platforms, along with a robotic arm assembly and its in-house CHAMP joint actuator modules.
The display offered attendees a rare opportunity to engage with hardware and software components developed end-to-end within a single organization, from individual actuators to fully autonomous robotic systems.
GENISOM M1: Industrial-Grade Capability at the Core of the Showcase
The GENISOM M1 quadruped robot was the primary product spotlight at ICRA 2026. Positioned as the industry’s first lightweight, high-payload, fully protected quadruped robot, the M1 achieves a 1:1 load-to-weight ratio, supporting a continuous walking payload of 30kg.
It carries an IP67 ingress protection rating, operates across a temperature range of -20 to 55 degrees Celsius, reaches a maximum wheeled speed of 8m/s, and clears obstacles up to 80cm in height at a maximum 45-degree climb angle.
The GENISOM M1 is powered by GENISOM AI’s in-house CHAMP P85MAX-S joint actuator modules, which deliver a peak torque of 180N·m. Dual hot-swappable batteries provide up to five hours of operation under no load, and 15 open interfaces support integration with external sensors, power systems, and control hardware.
These specifications make the GENISOM M1 a practical candidate for power grid inspection, industrial facility monitoring, emergency response, logistics, and security patrol. These are environments that demand sustained autonomous operation in conditions too hazardous or physically inaccessible for conventional ground platforms.
A Complete AI Stack: From Simulation to Agentic Intelligence
While the GENISOM M1 served as the hardware centerpiece, GENISOM AI’s ICRA presentation also extended across the software and algorithmic layers that make autonomous robot operation possible at scale.
The MATRiX simulation platform was one of the most significant technical highlights at the event. An open-source, next-generation simulation environment, MATRiX combines the MuJoCo physics engine with Unreal Engine 5 for photorealistic rendering, integrated with native ROS2 interfaces for multi-sensor data streaming and algorithm interaction.
Built on the REAL2SIM2REAL data flywheel, MATRiX supports seamless reuse of scene assets—from 3D Gaussian Splatting and generative Gaussian scenes to point clouds and mesh models—across the entire simulation workflow. This significantly reduces hardware trial-and-error costs and shortens the time from simulation to physical deployment. Developers can access MATRiX at https://github.com/zsibot/matrix.
The RoamerX intelligent navigation platform handles autonomous mobility across the robot lineup. It integrates real-time LiDAR, IMU, and visual sensor mapping, centimeter-level localization through a visual-LiDAR descriptor model with drift compensation, and a Spatio-Temporal Planning Algorithm that generates efficient paths in dynamic environments. The result is stable autonomous navigation validated in both simulation and public real-world demonstrations.
At the motion control layer, GENISOM AI’s Whole-Body Control framework enables quadruped-manipulator robots to simultaneously track body velocity and end-effector pose. The policy is trained entirely in simulation through reinforcement learning and imitation learning, then transferred to physical hardware without any additional real-world data collection.
Completing the stack is SomaMind, GENISOM AI’s agentic system for physical and interactive intelligence. SomaMind connects high-level AI reasoning with physical robot execution through a skill orchestration layer built on behavior trees.
It decomposes complex user instructions into sequences of atomic skills covering navigation, perception, manipulation, and communication, and allows developers to extend the system by writing reusable skills on top of existing robot capability modules.
Real-World Deployment and Production at Scale
GENISOM AI’s technology has already been deployed in operational settings. Current deployments span power grid inspection, security patrol, emergency rescue, logistics and transportation, industrial workshop monitoring, and education and research programs.
Underpinning this deployment capability is an established manufacturing operation. GENISOM AI has reached a milestone of 10,000 units manufactured and delivered across its quadruped robot platforms, evidence of its readiness to support production-grade robotics programs. The company’s facility in Suzhou, China, operates automated assembly lines under ISO 9001, ISO 45001, and ISO 14001 certified quality management systems.
The company’s hardware has also been independently validated in competitive conditions. A team from the University of Manchester used GENISOM AI hardware to claim first place at the IROS 2025 Quadruped Robot Challenge, one of the most rigorous benchmarks in legged robotics, confirming the reliability and performance of the company’s robotic systems under pressure.
Learn more information about GENISOM AI at http://www.genisomai.com.
About GENISOM AI
Founded in December 2023 and headquartered in Beijing, GENISOM AI is a full-stack embodied robotics company with capabilities spanning core components, robot platforms, manufacturing, simulation, autonomous navigation, motion control, and AI agent systems.
The company develops quadruped and humanoid robot platforms while also supporting partners across the robotics value chain through in-house joint actuator modules, platform customization, ODM and production support, software toolchains, and secondary-development resources. Its technologies are designed to help system integrators, robotics companies, research teams, and industry partners accelerate robot development and real-world deployment.
GENISOM AI has approximately 400 employees, with around 70 percent dedicated to research and development. The company has surpassed 10,000 units in cumulative production and delivery across its quadruped robot platforms, with applications spanning industrial inspection, security patrol, emergency response, and research and education.
