At ICRA 2026, GENISOM AI could have been new to many worldwide attendees — however it’s not a concept-stage robotics startup. Based in December 2023, the corporate stated it has already produced and delivered greater than 10,000 models, making it one of many few robotics firms to succeed in manufacturing scale in beneath three years.
GENISOM AI stated that milestone places it in a unique place than many early-stage bodily AI firms coming into the worldwide market. Like Unitree, GENISOM AI is constructing round manufacturable platforms designed to maneuver past laboratory demonstrations and attain clients at scale.
Nevertheless, the Beijing-based firm stated it places better emphasis on trade deployment, combining robotics {hardware} and software program, in-house core applied sciences, and real-world utility capabilities.
M1, the corporate’s industrial-grade quadruped robotic
GENISOM AI builds platforms for real-world deployment
At ICRA in Vienna final week, GENISOM AI showcased its M1 and L1-series robots as mature, mass-produced methods.
The GENISOM M1 is the corporate’s industrial-grade quadruped robotic, rated for a 30 kg (66.1 lb.) steady strolling payload and a payload-to-weight ratio approaching 1:1. The platform additionally carries an IP67 safety ranking and might help as much as 5 hours of runtime, relying on payload and working situations.
That payload capability is supported partly by GENISOM AI’s in-house P85MAX-S joint actuator module, which delivers as much as 180 N·m peak torque in an 86 mm (3.3 in.) diameter type issue weighing roughly 1 kg (2.2. lb.). The corporate stated that creating actuators in-house offers it direct management over a key hardware-software integration level that shapes locomotion efficiency, payload capability, and system reliability.
The GENISOM M1 Extremely provides a sophisticated notion layer to the sequence. It attracts on chook’s eye view (BEV)-based temporal fusion and occupancy-network strategies — approaches extra generally related to autonomous driving notion. This strategy helps the Omni-Panorama system for 720° 3D spatial consciousness and a extra semantic understanding of surrounding constructions, obstacles, and spatial relationships.
Past the present ground, GENISOM AI stated its robotic platforms have already moved into real-world purposes, together with analysis and schooling, safety patrols, and emergency response, with completely different platforms serving completely different wants.
Instructional platform wins first place at IROS
On the analysis and growth aspect, the GENISOM L1 EDU combines NVIDIA Jetson Orin NX computing with Livox Mid-360 lidar, RealSense depth cameras, GNSS, and 5G connectivity. It’s designed to help simultaneous localization and mapping (SLAM), 3D reconstruction, autonomous navigation analysis, and open-source growth workflows.
Notably, a College of Manchester group won first place on the IROS 2025 Quadruped Robotic Problem utilizing a GENISOM L1 EDU platform, additional validating its mobility and navigation efficiency.
Full-stack know-how permits M1 and L1
Behind the M1 and L1-series platforms is a set of in-house software program, simulation, navigation, and management applied sciences that help them throughout coaching, growth, and deployment.
MATRiX, GENISOM AI’s open-source simulation platform, combines the MuJoCo physics engine with Unreal Engine 5 rendering to help reinforcement studying coaching and sensor simulation throughout RGB, depth, lidar, IMU, and panoramic inputs. The corporate stated it’s designed to handle a longstanding robotics simulation problem: combining bodily accuracy and visible realism inside a single setting.
Working alongside MATRiX is the Real2Sim2Real Knowledge Flywheel, which converts real-world environments captured with consumer-grade cameras into training-ready simulation property, reducing the price of coverage growth and validation.
On the autonomy layer, RoamerX helps mapping, localization, path planning, impediment avoidance, and cross-floor navigation. The Complete-Physique Management framework coordinates legged locomotion and manipulator movement for cellular manipulators.
SomaMind, the corporate’s Bodily AI Agent System, connects high-level activity understanding with bodily execution modules.
Companions get help throughout the robotics stack
GENISOM AI stated its mannequin extends past promoting particular person robotic models. Its capabilities span core parts, robotic platforms, manufacturing capabilities, software program toolchains, and algorithms, permitting companions to work with the corporate at completely different layers of the robotics stack.
For system integrators, GENISOM AI supplies platforms and help for area purposes resembling inspection, safety patrol, emergency response, and vitality operations. Robotics manufacturers and firms creating their very own robotic merchandise can work with GENISOM AI on ODM (unique design producer) customization and manufacturing enablement.
Analysis and secondary-development groups can construct on open interfaces, a software program growth equipment (SDK), and RoamerX navigation software program resources obtainable on GitHub.
By combining robotic {hardware}, manufacturing capabilities, software program instruments, and field-application expertise, GENISOM AI stated it’s constructing a broader robotics enterprise — one which goes past particular person product gross sales.
GENISOM AI described its ICRA showcase as greater than a primary look at a significant robotics convention. Reasonably than centering on a single product launch, the showcase positioned GENISOM AI as a robotics platform firm targeted on manufacturing, growth, and real-world deployment.
Sponsored content material by Genisom AI
The publish GENISOM AI debuts deployable robotics platforms at ICRA 2026 appeared first on The Robotic Report.
