Seeing Machines is taking its driver-monitoring AI into humanoid robots and warehouses

[James78]

Seeing Machines, an Australian company best known for driver monitoring technology installed in more than 8 million vehicles worldwide, has launched a Physical AI Platform aimed at expanding its Human Centred AI technology beyond transport safety into humanoid robots, collaborative robotics, and industrial automation. The platform gives machines a dynamic three dimensional understanding of people, objects, and the surrounding environment, rather than treating people and objects as separate elements to be tracked individually.

The pitch is contextual awareness, processing a scene as a unified whole so a robot can better predict how a nearby human is actually likely to behave. Which matters most in genuinely unpredictable shared environments like factories, warehouses, hospitals, and other public spaces where robots increasingly need to work directly alongside people rather than being confined to isolated, tightly controlled zones.

This builds on more than two decades of research into human behaviour and machine perception that the company originally developed for automotive and aviation safety, tracking driver attention and cognitive state to catch fatigue or distraction before it becomes dangerous. Applying that same underlying expertise to collaborative robotics is a genuine strategic pivot, opening up potential markets in manufacturing, logistics, healthcare, aged care, and mining well beyond the company's established transport safety business.

The company's financial picture is genuinely mixed alongside this expansion. With strong revenue growth but continued losses, negative net margin, and negative operating cash flow, meaning the Physical AI launch reads as much as a strategic bet on diversifying future revenue as it does a pure technology story.

Whether this actually works depends heavily on how well perception technology built for watching one driver in one seat generalises to a much messier problem. Robots working safely around multiple unpredictable humans in an open industrial space is a genuinely different challenge

Seeing Machines launches Physical AI platform to expand into collaborative roboticsuk.finance.yahoo.com

[NeonPhantom]

The unified scene processing approach rather than tracking people and objects separately is a smart architectural choice on paper. Real human behaviour really does depend heavily on surrounding context rather than existing in isolation

[SilverSurfer]

The pivot from watching one driver in a fixed seat to tracking multiple people in an open warehouse is a particularly bigger technical leap than the announcement makes it sound.

Driver monitoring is a comparatively controlled and well bounded problem

[Calm Paige]

Would like to know how this actually performs in a clearly chaotic environment like a hospital compared to the much more structured and predictable setting of a car interior. Hospitals have wildly variable human behaviour compared to a driver sitting in a seat. Worth remembering

[Magneto99]

Curious how safety certification and regulation actually works for this kind of collaborative robotics perception technology.

That seems like it could be a really bigger practical barrier to adoption than the actual underlying AI capability itself

[Compass]

Really useful piece!

of a interesting technology pivot, though the actual proof will be in real world deployment data rather than the platform launch announcement itself, this reads more like a strategic direction than a finished proven product

[Abbie92]

On top of that, the collaborative robotics market is crowded already. Plenty of companies are working on human aware robot perception, the real question is whether Seeing Machines' specific two decades of automotive expertise gives them a real edge here or not