A blind-assistant robot backed by state and industry capital shows where embodied intelligence is finding paying demand: in tough, real human need.
OFweek reported on 16 October that Shandong Zhuge Robotics closed a Pre-A round with Pudong Venture Capital, Bawei Tong Spatial Intelligence and Shijiazhuang Quchuang Technology together investing. At a moment when most embodied-AI projects still sit in lab demos, a robot for visually impaired people drew both industrial and state-backed capital, a sign of a structural shift in the sector toward commercial scenes with real social value.
Why capital took notice
Unlike many flashy humanoid startups, Zhuge anchored from day one on the real, stubborn pain of travel for the blind, a path stamped by founder Zhu Qingyi’s own life. A fever at 15 dropped his vision to 0.01. He studied at Changchun University of Special Education, taught himself to code, built China’s first accessible website in 2001, and for over twenty years helped write national and international accessibility standards as a recognised expert for the China Disabled Persons’ Federation. Because he is himself part of the blind community, he understands the small, fatal obstacles on a route better than any outside team, and pushed the company from software accessibility toward hardware robots.
What makes the robot hard
Many treat a blind-assistant robot as a bigger navigation device, but real streets are far messier than phone maps. The AI LOOK robot breaks from the pre-scan-map and remote-control pattern. Rivals need staff to map an area first, and any change breaks the system, useless on ever-shifting city pavements. Zhuge’s robot uses multi-sensor fusion and a pavement vision large model to perceive, decide and guide in a strange public space directly, judging pedestrians, obstacles and passable areas, and predicting risk ahead. The team balanced compute heat and outdoor waterproofing through repeated real-device tests. The moat is the real-scene data: lab simulation can never replace true street conditions, and every trip retrains the model.
The signal from this round reaches past one startup’s cheque. For a long stretch, Chinese embodied AI chased general humanoids and cool demos, but high cost and missing scenes blocked commercial use. Accessibility offers another route: a rigid, specific need that scales. Public spaces, campuses and aged-care markets open a commercial imagination, though cost, standards and scale remain hurdles. After this round, Zhuge plans to keep deepening its model, harden its scene-data moat, and broaden deployment across the country. When AI built by the person who lives the pain enters daily life, that may be the future of embodied intelligence worth seeing.
Editor’s note: This is an adapted translation of the original OFweek report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://robot.ofweek.com/2026-10/ART-8321203-8420-30707450.html.