Wireby Fusion42
Read this story on the live Wire →

Wire · founder news, decoded · technology

Glass crashes slashed? Ant Group embodied AI unit claims breakthrough in robot sensing

Ant Group's embodied AI unit Robbyant unveiled LingBot-Depth 2.0 and LingBot-Vision, foundational models that solve critical robotics perception challenges—particularly accurate sensing of glass, mirrors, and transparent objects—while achieving superior performance using 7x fewer parameters than Meta's DINOv3.

This Wire brief sits within Fusion42's coverage of AI Agents, Robotics and Computer Vision. Wire is Fusion42's founder-focused intelligence feed: each story is connected to the funds and startups it names — every one with a live profile on Raise or Scout — so founders can follow the capital and the momentum behind the headline rather than just the headline itself. Wire analysis is one of the live surfaces Arthur, Fusion42's AI co-founder, reasons over.

The Wire takeaway

Founders building robotics, autonomous systems, or embodied AI should monitor this—efficient, open-source foundational vision models with solved glass/transparency sensing directly reduce perception costs and timelines for hardware deployment.

Read the full story at amp.scmp.com

Topics: AI Agents · Robotics · Computer Vision · embodied-ai · robotics-perception · vision-models · depth-estimation · foundational-models

Related on Wire

Verified 8 July 2026 · Sources: Fusion42 review