Striding AI unveiled two store-floor robots and an orchestration platform in Singapore on October 8, 2026, targeting autonomous overnight restocking and customer service for 24-hour retail chains.
Founder and CEO Song Yao introduced the setup at APRCE 2026, where the company reported the system completes up to 99 percent of retail tasks autonomously.
The suite combines the C1 mobile manipulator with the H1 humanoid robot, designed to automate shelf replenishment, inventory checks, carrying, and patrols while operating in narrow aisles without extensive store modifications.
The C1 mobile manipulator navigates tight spaces on an omnidirectional chassis with an operational reach spanning 100 mm to over 1,750 mm. Front and rear depth sensors, 360-degree vision, and emergency braking for obstacles detected within 30 cm support safe navigation around shoppers.
Working alongside it, the H1 humanoid robot handles customer greetings and guidance using multimodal interaction across more than 20 languages, 45 facial expressions, and 36 gestures, supported by open APIs for retailer integration.
Both machines operate through Striding AI’s SLM-0.5 latent world action model, which computes decisions in latent space rather than relying on pixel-by-pixel prediction. Under reported evaluation settings, the system achieves 187 ms latency per action chunk, delivering a 98.6% success rate on LIBERO and 91.22% on RoboTwin.
Real-world operations feed back into a reinforcement learning framework called STEAM, designed to refine operational policies from physical floor experience. Coordinating the fleet is the M1 management platform, which uses edge autonomy so core navigation, task execution, and grasping continue without continuous network connectivity.
Source link
