Robotics & Physical AI▲
Seeing Machines Publishes Technical Paper on Real-Time 3D Human Understanding for Physical AI
Seeing Machines Limited published a Technical Paper demonstrating how robots and other Physical AI systems can achieve real-time 3D human understanding directly on embedded hardware. The paper presents the company's Human Mesh Recovery technology, a core capability of its Human-Centred Physical AI Platform launched in August 2026, which uses a single camera view to reconstruct a detailed 3D representation of a person so a machine can understand their location, posture and movement in real time. Deployed on NVIDIA Jetson Thor hardware, the technology delivers up to 180 FPS while combining state-of-the-art accuracy with low latency and compact model size, and the HMR model supports both standard RGB cameras and RGB-D cameras without increasing its parameter count. CEO Paul McGlone said the technology brings detailed 3D human understanding into real-world environments such as factories, hospitals and warehouses, drawing on Seeing Machines' 25 years of human-sensing expertise. Seeing Machines, founded in 2000 and headquartered in Australia, operates across Australia, the USA, Europe and Asia.
About megatrends
Robotics & Physical AI › Machine Vision & Force/Tactile Sensing ▲Technology
Robotics & Physical AI › Robotics AI & Embodiment Software ▲Technology
SEE.LSE · Technology · Positive Seeing Machines published a technical paper demonstrating its Human Mesh Recovery technology for real-time 3D human understanding on embedded hardware.
NVDA · Demand · Positive Seeing Machines' HMR technology is deployed on NVIDIA Jetson Thor hardware, indicating adoption of NVIDIA's embedded platform.