Seeing Machines
About
About Seeing Machines
Seeing Machines began in 2000 as an Australian National University spin-out focused on non-contact face and eye tracking. Its transport products combine human-factors research, optical design, embedded processing and AI algorithms to estimate gaze, eyelid movement, head pose and operator state in real time. The technology is used in automotive, fleet, off-road and aviation safety systems.
The Physical AI Platform extends that work into humanoid and industrial robotics. It builds a dynamic three-dimensional map of people, objects and their environment, then interprets spatial relationships and human behaviour so a robot can anticipate risk and adjust its actions around people.
Observed use
- Jetson Thor
Research Experiment - Tested With - HMR and HMR-Lite ran on NVIDIA Jetson Thor for embedded human-mesh benchmarks; HMR-Lite’s 5.5 ms RGB latency excludes person detection.
Source
Team
- Paul McGlone - Chief Executive Officer - Leads Seeing Machines as it extends its human-centred perception technology into robotics and factory automation.
- John Noble - Chief Technology Officer - Leads technology development for Seeing Machines, including its 3D perception-mapping architecture.
Research
- Seeing Machines publishes Jetson Thor human-mesh recovery benchmarks
2026-09-30 - Research Publication - HMR-Lite achieved 5.5 ms median RGB latency and 6.2 ms for RGB-D; person bounding-box detection time is excluded. The larger HMR model achieved 15.8 ms RGB latency with 386 million parameters, versus 146 million for HMR-Lite. Both configurations reconstruct body pose, shape and camera-frame placement from a single RGB or RGB-D view.
Source
Timeline
Events
- 2026-09-30 - Research Publication - Seeing Machines publishes Jetson Thor human-mesh recovery benchmarks
HMR-Lite achieved 5.5 ms median RGB latency and 6.2 ms for RGB-D; person bounding-box detection time is excluded. The larger HMR model achieved 15.8 ms RGB latency with 386 million parameters, versus 146 million for HMR-Lite. Both configurations reconstruct body pose, shape and camera-frame placement from a single RGB or RGB-D view.
Source - 2026-08-17 - Product Launch - Seeing Machines launches Physical AI Platform for human-aware robotics
The platform builds a dynamic 3D map of people, objects and their environment. It interprets spatial relationships and human behaviour to support real-time risk anticipation and robot decision-making. Target applications include manufacturing, logistics, healthcare, care services, warehousing, mining and industrial automation.
Source - 2026-08-12 - Contract Award - Seeing Machines secures advanced robotics development contract
The initial programme applies Seeing Machines' Perception Map and Human-Centred AI to a robotics proof of concept in a factory-automation environment. The companies expect the programme to establish a basis for further collaborative development across industrial automation and robotics.
Source
Commercial
- Seeing Machines secures advanced robotics development contract
2026-08-12 · Order / contract · Seeing Machines secured a development contract with an unnamed global industrial-technology company for a factory-automation robotics proof of concept.
Source