MenloAI presents compact Asimov 2 humanoid
English translation from Unibot. Source published 2026-08-13.
MenloAI presented Asimov 2, a compact humanoid for developers, researchers and education teams. It is a practical physical platform rather than a concept machine.
Specifications and architecture
The robot stands 120 centimeters tall, weighs 38 kilograms and has a maximum speed of 4 kilometers per hour. Its stated payload is 5 kilograms and runtime is 2.5 hours per charge. The robot is safe around people.
Its hybrid architecture assigns media and networking to Raspberry Pi 5 and movement control to Radxa CM5. This separation of computing is useful for experiments and practical work.
Connectivity includes internal CAN buses, Bluetooth, Ethernet and Wi-Fi. Software runs on Asimov OS with LLM integration.
The overall skill score is 2, with navigation and manipulation each rated 1. This is an early stage focused on basic movement and simple tasks rather than complex whole-body actions.
What Asimov 2 can do
The robot focuses on two specific tasks: obstacle avoidance and goods sorting. Obstacle avoidance provides basic spatial awareness, allowing the platform to move confidently through cluttered rooms among people, furniture and working objects. It goes beyond a scripted demonstration robot to a system with elements of control.
Goods sorting targets practical scenarios in laboratories, classrooms and light logistics settings, where repeated object selection and placement matter more than speed. Both scenarios match early navigation and manipulation levels, emphasizing learning, integration and iteration.
For teams building perception, control and task-planning pipelines, Asimov 2 offers a way to connect sensing, motion control and simple workflows in one embodied system. It can be tested, refined and evaluated against metrics over an extended period.
Price and positioning
Asimov 2 is available as a prototype. It forms part of MenloAI's wider strategy to make humanoid robotics more accessible, expanding from Singapore into global developer and education markets.
The price is US$35,000. For universities, robotics programs and ambitious laboratories, it is a platform for long-term experiments rather than a one-off exhibit.
The model's value lies in combining an open-platform philosophy with a complete humanoid body. Teams can study locomotion, manipulation, onboard computing and safe operation near people in one system.
This makes Asimov 2 relevant despite its modest current scores: it supports everyday software development, behavior testing and the education of a new generation of roboticists using hardware that can actually be put to work.