Asimov has released a Python software development kit for applications to read its humanoid’s state and command movements. The Apache 2.0 release includes walking, joint control and camera and audio access. A separately announced Edge layer is intended to arbitrate between control sources; its source availability remains unconfirmed, and the SDK is explicitly not an emergency stop.
Asimov has released a Python software development kit (SDK) for its Asimov 1 humanoid. The public Menlo SDK library gives applications a way to read the robot's state and send movement commands.[1], [2]
The release extends the project's application software after earlier hardware designs and locomotion-training code. Developers can use high-level walking commands or direct joint control, with access to the camera, microphone and speaker.[1]
The repository requires Python 3.12 or newer and uses the Apache 2.0 licence. LiveKit, a connection system used for remote teleoperation, provides the internet path; a UDP interface for sending data directly is described for local applications. A team demonstration sends robot state into a walking policy and returns its output as joint commands. The team says that computation can run on a laptop or the robot's onboard Pi computer.[1]
Asimov also announced opening Edge, a layer that manages competing human and application control sources before forwarding commands to firmware. Human takeover from an automated application is one example. The existing documentation gives Cockpit and a paired gamepad priority over the SDK, which it explicitly identifies as not being an emergency stop. The SDK repository is available, while Edge source availability remains unconfirmed. The team demonstrations do not establish independent field-safety results.[1]
