Highlights of the First Didi Industry-Academia Collaboration Forum and Gaia Lighthouse Project Awards (June 2023)

On June 21, 2023, the first Didi Industry‑Academia Collaboration Forum and the Gaia Lighthouse Project 2021‑2022 Outstanding Project Award ceremony were held in Beijing.

The forum, organized by Didi’s CTO and autonomous‑driving CEO Zhang Bo and other senior leaders, brought together Didi executives, academic partners from Guangdong University of Technology, Beijing Jiaotong University, Tsinghua University and Tianjin University, and highlighted the goals of Didi’s industry‑academia program: to combine frontier academic research with real‑world business scenarios, jointly define problems, and translate scientific innovation into industrial value.

Awarded Projects

GPS‑Missing Scene Perception and Position Tracking Team: Associate Professor Gao Ruipeng (Beijing Jiaotong University) and Didi Map team. The project deployed a deep vehicle‑position inference model on‑device to improve positioning in weak‑signal environments such as long tunnels, achieving a 17 % performance gain over competitors and producing nine academic papers (five SCI).

Long‑Term Effect Estimation Based on Short‑Term Proxy Variables Team: Professor Cai Ruichu (Guangdong University of Technology) and Didi ride‑hailing team. A new model extracts effective representations from massive short‑term observable factors to evaluate long‑term pricing‑strategy effects, incorporating domain‑adaptation across cities and delivering over 50 % relative improvement in long‑term effect prediction and more than 70 % improvement in city‑transfer accuracy.

High‑Precision Positioning for Two‑Wheeled Vehicles Team: Professor Liu Hui (Wuhan University) and Didi two‑wheel team. The collaboration produced a centimeter‑level positioning algorithm for complex urban canyon scenarios, completing three development phases and integrating the system into the two‑wheel business.

In‑Vehicle Voice Separation Team: Professor Wang Longbiao (Tianjin University) and Didi intelligent voice interaction team. The joint research addressed multi‑speaker interference in cars, delivering a four‑zone voice‑separation algorithm with robust performance and two conference papers.

Data‑Driven Decision‑Control Technology for High‑Level Autonomous Driving Team: Professor Li Shengbo (Tsinghua University) and Didi autonomous‑driving team. The project built a full data‑driven decision‑control solution, completed the first brain‑inspired learning autonomous‑driving system on real vehicles, released two core training/simulation tools, published multiple top‑conference papers (one winning best paper), and cultivated several graduate students.

In his opening speech, Zhang Bo emphasized that the next decade will be driven by general artificial intelligence, which will greatly upgrade Didi’s platform, improve safety and efficiency, and create new value through deeper collaborations with academia.