Didi Chuxing is the largest ride-sharing company in China, and a leader in China on Artificial Intelligence and Autonomous Technology.  U-M has collaborated with DiDi since 2017, and is DiDi’s largest academic research partner. Through MIDAS coordination, DiDi has funded 19 U-M data science projects in the past three years, involving faculty members in the College of Engineering, College of Literature, Science and the Arts, Ford School of Public Policy, Ross School of Business, School for Environment and Sustainability, School of Information, School of Public Health, and UM Transportation Research Institute.  

DiDi collaboration focus:
The focus of DiDi funded projects includes developing data science methodology, using data science to improve business operations, building a fleet of electric vehicles, improving passenger and driver experience, and making a social impact.

Spotlight projects

Making Machine Learning more Robust and Trustworthy

Led by Dr. Atul Prakash (Electrical Engineering and Computer Science), this group develops strategies improve the robustness of machine learning models against small perturbations.  In practice, such strategies can improve the recognition of traffic signs even if they are dirty or defective, thus improving the performance of autonated vehicles. This work also helps us gain a deeper understanding of why deep learning models behave incorrectly when exposed to unanticipated or adversarial inputs. 

Improving Driver Performance through Team Competition

Led by Dr. Yan Chen and Qiaozhu Mei (School of Information), this research group has worked with DiDi for three years to help drivers form teams with compatible partners. While workers in the sharing economy enjoy autonomy and flexibility, a lack of identity through work, knowledge sharing and a career path can lead to low job satisfaction, slow skill improvement, less engagement and high attrition.  This research group’s approach is to use a team competition and reward system to facilitate bonding among drivers, knowledge sharing, hence increase driver engagement and job satisfaction. Their work has already been adopted in many Chinese cities.

Improving Didi’s Operations via Enhanced Matching, Repositioning and Contract Design

Dr. Yafeng Yin (Civil and Environment Engineering) and Dr. Xiuli Chao (Industrial and Operations Engineering) develop advanced matching and repositioning techniques which can improve the efficiency of Didi’s ride‐hailing system.  They analyze empirical data, develop models and optimize the system in real time.