Menlo Research’s Asimov v1 pushes humanoid teleoperation into open-source territory

A compact humanoid appears

Humanoid robots are still being shaped around a hard set of problems: balancing on two legs, coordinating whole-body motion, and making manipulation feel responsive in human spaces. Menlo Research has now brought Asimov v1 into that conversation as an open humanoid kit, with official documentation and third-party coverage describing a 1.2-meter, 35-kilogram biped designed for research, demos, and teleoperation-first development.

Why it stands out

What makes Asimov v1 notable is not a single high-end spec, but the way its design centers on full-body coordination rather than isolated arm motion. The platform combines visual sensing, balance correction, and simulation-backed locomotion in a form factor intended for human-centric environments, while its open hardware and software stack lowers the barrier for labs that want to test teleoperation, imitation, and manipulation workflows. It also arrives as the industry continues to favor humanoids that can be operated and trained in real time instead of relying only on pre-scripted motion libraries. Asimov v1 is less about a polished consumer robot than about making human-guided humanoid control easier to study.

Motion from input

Menlo’s technical flow for Asimov v1 is straightforward: human motion input is processed by the robot’s AI and control stack, which then drives joint actuation while balance corrections keep the biped upright. The official documentation points to visual SLAM and MuJoCo-based simulation as core parts of that pipeline, with ROS2, Python, C++, and MuJoCo forming the software layer that connects perception, planning, and locomotion. In practical terms, the system is built to translate what a person does or demonstrates into controlled whole-body movement.

Remote task practice

A realistic deployment scenario is remote inspection or assistance inside a human environment, where a teleoperator needs a robot that can move through standard spaces, handle tools, and interact with objects at close range. Asimov v1’s compact bipedal build and reported payload focus on tools, packages, precision instruments, and person-to-person interaction make it better suited to that kind of structured indoor work than to rough outdoor labor. That focus matters because the category is increasingly being evaluated on how naturally a human can guide the robot, not just on whether it can walk.

What the hardware enables

The reported 120 cm x 45 cm x 30 cm footprint and 35 kg weight make Asimov v1 a comparatively manageable humanoid for labs that need to move and set up hardware without industrial infrastructure. Its sensor package, including a 2MP monocular RGB camera, four microphones, an IMU, gyroscope, force sensors in the joints, and temperature sensors, supports the kind of perception and feedback loop needed for balance, teleoperation, and interaction experiments. Menlo’s official specs also describe ROS2, Python, C++, and MuJoCo as the software backbone, which makes the platform more accessible to robotics teams already working in those ecosystems.

Rivals Edge Check

The market signal

Asimov v1 points to a more practical phase in humanoid robotics, where the near-term value proposition is not full autonomy but reliable human-in-the-loop control in spaces built for people. That shift matters because it favors platforms that can be taught, inspected, and iterated on quickly by researchers and integrators, rather than systems that depend on closed motion libraries and tightly managed deployments. If that trend holds, the next round of humanoid competition will be decided as much by software openness and teleoperation fidelity as by mechanical design.