Localization with sensor fusion can integrate information from multiple sensors such as GNSS, IMU, LiDAR, vision, and wheel odometry to take advantage of the complementary functionality of different sensors. This enables high-precision and high-reliability positioning information in different scenarios, particularly with support for zero-speed correction. Unique accurate relative positioning can realize positioning in long tunnels, GPS-denied areas without satellite signals, among other scenarios. CiDi has deployed commercial systems with GPS-free autonomous driving in scenarios 1,000 meters underground.

Multimodal heterogeneous perception fusion algorithms combining different types of sensors can implement 360-degree, blind-spot free perception. Adaptive fusion of V2X roadside perception information achieves over-the-horizon capabilities, and intelligent switching of sensor fusion copes with various working conditions and faults during operation. High-precision and self-calibration and algorithms can provide precise time-space synchronization of sensors, enabling the efficiency required by mass production or accurate long-distance perception systems. At high-speeds, stable perception distance is more than 300m, with a maximum distance of 1500m, conditions permitting. Lane-level obstacle information above 1000m is achievable without relying on mapping.

Through cutting-edge predictive control and planning technology, CiDi achieves precise and stable control of autonomous heavy duty commercial and industrial vehicles in different scenarios. To ensure adequate robustness and the ability to quickly deploy to different vehicle types, high-fidelity vehicle simulations are used to automatically and systematically identify complicated chassis and vehicle systems, including systems with hydraulic steering, braking, and various types of powertrains. We have cooperated with major domestic OEMs including DFLZ, Shaanxi Automobile Group and Foton Daimler on heavy road trucks and ultra-class mining haul trucks.