Geely is developing an All-Terrain 2.0 system that currently achieves road surface recognition accuracy of over 95% with a response time under one second. Yang Zijian, a senior chief engineer at the Geely Auto Research Institute’s Intelligent Architecture Center, revealed the progress at The 6th Automotive Al Intelligent Chassis Conference 2026. Unlike the previous 1.0 version, which relied primarily on cameras and driver-assistance systems, the 2.0 iteration fuses chassis sensor data and leverages AI for feature selection and recognition modeling. The upgrade expands the range of detectable road types and enhances precision, while automatically matching driving modes to reduce the need for manual adjustments. Yang highlighted that Geely, working with its European R&D teams, has completed real-world demonstrations and extensive mileage testing for notoriously difficult surfaces like damaged roads and Belgian pavement. The recognition capabilities for these scenarios are undergoing continuous optimization, though the All-Terrain 2.0 system remains in the development phase. On the chassis architecture front, Geely has achieved coordinated control across longitudinal, lateral, and vertical dynamics through its GVMC vehicle motion control system. Combined with high-precision road preview data from driver-assistance systems, this enables features like the "Super Magic Carpet" ride. Geely is advancing cross-domain integration that links the chassis with driver-assistance and smart cockpit systems, Yang said. This includes full-scenario road recognition, driver style profiling, and preventive collision control for sudden hazards like "ghost probes"—where pedestrians or vehicles emerge unexpectedly from blind spots. Notably, an AI-based hands-off detection function for the steering wheel has already been implemented on the chassis side. Beyond road driving, Geely is researching track racing capabilities and braking distance reduction through the collaboration of driver-assistance and chassis systems, aiming to replicate the performance of professional race car drivers on the track. Regarding the role of AI in chassis systems, Yang argued that while AI is better suited for complex environmental perception and personalized decision-making, safety-critical real-time closed-loop controls—such as steering and braking—must still rely primarily on rule-based algorithms. The goal is a complementary approach where AI enhances, rather than replaces, traditional control technologies.