winning the championship of mcity global autonomous driving challenge, tsinghua yang diange team delivers another good news
2024-09-30
한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina
the picture shows the score list of the mcity global autonomous driving challenge. photo provided by interviewee
china youth daily client news (china youth daily·china youth daily reporter zhang zhenqi) as autonomous driving has become a technological highland where the world's major technological powers are competing to deploy, various technical challenges have become an arena for domestic and foreign technical masters to compete. at the just-concluded mcity global autonomous driving challenge, the thu ad lab team from the research group of professor yang diange of the school of vehicle engineering of tsinghua university reported another good news, winning the competition with first place in total score and first in safety score.
as the world's top autonomous driving algorithm competition, the mcity global autonomous driving challenge is organized by the university of michigan, attracting top universities and car companies from the united states, china, australia and other countries to participate in the competition. at the award ceremony held at the international intelligent transportation conference in edmonton, canada, professor henry liu, director of mcity in the united states and university of michigan, ann arbor, presented the award to the team. team member zhou weitao, a postdoctoral fellow in the school of vehicles, was invited to report the technical solution on behalf of the team.
according to reports, this competition evaluates the driving capabilities of autonomous driving algorithms in real-life traffic environments by generating natural driving scenarios including long-tail scenarios (corner cases). the competition uses real road traffic in michigan, usa, from 2016 to 2021. accident collision data calibration and test environment generation. the autonomous driving algorithm participating in the competition needs to deal with long-tail scenarios (corner cases) during continuous driving to ensure safety, while taking into account driving evaluation indicators such as driving efficiency, comfort, and traffic rules.
an autonomous driving algorithm with autonomous risk response capabilities built by the thu ad lab team. photo provided by interviewee
in the competition, the thu ad lab team built an autonomous driving algorithm with autonomous risk response capabilities, and safely passed all long-tail scenarios (corner cases) during the competition, defeating mit, university of michigan, purdue university, and general motors. experts in the industry, including no. 1 in overall score and safety score, won the championship.
the reporter learned that in response to the long-tail challenges of autonomous driving in open road scenarios, the research group of professor yang diange of the school of vehicles at tsinghua university conducted a series of studies, and the relevant results were applied in the autonomous driving demonstration project of the 2022 beijing science and technology winter olympics. in addition, the research team’s preliminary research has received strong support from autonomous driving-related companies such as didi, saic, dongfeng, and toyota.
it is worth mentioning that the relevant person in charge of the research team revealed that they are currently seeking more opportunities for school-enterprise cooperation to further carry out large-scale driverless open road application demonstrations.
source: china youth daily client
