AGIBOT KuTuo, a quadruped robotics business under Chinese robotics company AGIBOT, showcased upgrades to its Lingyue and Lingtu quadruped robots at AGIBOT INNOVATION DAY 2026 at the Shanghai World Expo Center, focusing on autonomous terrain switching and semantic navigation.
The Lingyue all-terrain intelligent motion engine now offers an “automatic gear” capability. By combining reinforcement-learning-based motion control with perception-based motion control, the robot can identify terrain changes and switch gaits automatically. The upgrade is designed to reduce manual or preset mode switching across stairs, platforms, obstacles, and narrow passages, improving travel efficiency in complex environments and lowering barriers to secondary development.
The Lingtu all-domain autonomous navigation engine demonstrated vision-language navigation (VLN) capabilities. Combining a visual semantic large model with an embodied navigation large model, the system enables robots to interpret targets, directions, and spatial relationships in natural language and use visual perception and map data to locate targets. The company said this extends navigation from coordinate-based movement toward understanding business objectives, simplifying task configuration and development.
The two technologies were first unveiled at the World Artificial Intelligence Conference (WAIC) in July. AGIBOT KuTuo said its quadruped robots have already been deployed for security patrols and industrial inspections. The company will continue to focus on perception-based motion control and autonomous navigation, targeting complex operating environments and large-scale commercial deployment.
AGIBOT
Figure AI
UBTECH