Menlo Research has a public Python SDK for Asimov 1. Its checked changelog dates version 0.1.0 to 9 October; Humanoids Daily covered the SDK and Edge announcement on 12 October. These are different dates—a software version and an announcement report—and we do not collapse them into an invented single-day launch.
The repository confirms an Apache 2.0 license and a requirement for Python 3.12 or newer. Before opening control, the SDK checks model and Robot OS compatibility; the changelog specifies Robot OS 0.2.3 or newer for that change. This gives the story an inspectable technical artifact rather than only a demonstration video.
Connecting an application
The official documentation describes control of standing, balance, movement, joints and media streams. Commands pass through Asimov Edge. Cockpit and a paired gamepad outrank the SDK. The documentation also warns that the SDK is not an emergency stop. The ability to send a command should not be confused with a guarantee that it is appropriate for the physical environment.
The API overview describes a robot with state and separate control sources. Our editorial reading is that this is an important architectural question in robotics: who currently has the right to command, how control is yielded and how the result becomes visible. Adding another model or client does not solve a conflict between two active command sources.
What remains unconfirmed
Humanoids Daily reports an intention to open Edge but explicitly says its newsroom had not found the source at review time. SIExpo confirmed the public SDK and documentation but did not establish a published Edge source release. We therefore do not call the complete control stack already open. The original X post was unavailable for direct reading during this check and is not used as an independently verified source.
Safety and engineering responsibility
The safety guide discusses checks before movement and dependence on correctly supported hardware and an operator. It is necessary context, not a safety certificate. This article is not an instruction to start a physical robot without preparation and contains no movement code.
Our editorial example is a laboratory workflow that first observes state, then determines the authorized control source and finally checks whether the command was executed. If communication fails, the system should make clear what is known and what remains uncertain. Sent, accepted and physically executed are separate events; a useful interface needs to expose that difference.
This distinction also limits what can be inferred from open code. A readable client helps a developer understand the requested operation, but does not establish the reliability of sensors, actuators or the physical environment. Those require their own evidence and tests. We do not infer such results from a software repository.
The SDK makes application development and inspection more accessible, but does not automatically make the robot a universal autonomous worker. SIExpo has not operated Asimov or measured its physical capabilities. We report the available code, the documented architecture and a clearly bounded promise about the next layer.
Sources checked on 13 October 2026. Availability and prices reflect the time of verification. The cover is an AI illustration, not a photograph of an actual event.
Editorial verification (UTC): 2026-10-13T06:10:05.979402+00:00
