29 youth teams from Russia, U.S.A., China, South Korea, France and Spain will take part in the IceVision competition final with the RUB 3 million prize pool. The contest aims to to look for AI-based solutions to enable driverless cars to recognize the traffic situation even under conditions of the Russian winter.
As part of an absentia selection stage that was completed on June 22, competitors tested their software to recognize traffic signs in the special winter data set that included 10,000 images shot from the car moving in winter conditions.
The teams that displayed the most efficient recognition algorithms were admitted to the final stage, including students from the Russian leading universities: the Moscow Institute of Physics and Technology (MIPT), Southern Federal University (SFEDU), Skoltech, Moscow State University (MSU), Novosibirsk State University, St. Petersburg Polytechnic University, National Research University Higher School of Economics, Innopolis University, etc. Representatives from the top foreign universities will compete with Russian developers. Teams from the University of Arizona, University of Wisconsin-Stout, Beijing Scientific and Technical University, Harbin Polytechnic University, Polytechnic University of Catalonia, etc. are invited to visit the IceVision finals in Moscow.
The final stage of the competition will be held on July 13-16 as a 48-hour hackathon at the site of the House of the Commune of NUST MISiS and Skoltech. The competitors will also be able to attend lectures and master classes, to meet developers of driverless cars participating in the UpGreat Winter City Contest and to ride driverless cars on the premises of the Skolkovo Innovation Center. StarLine NPO, a strategic partner of the contest, will conduct a special master class where the company’s developers will share their experience of creating a driverless car, the ways to draw up high-precision maps, localize cars in space, schedule the routes and the architecture of hardware and software solutions and will show their driverless cars in action. Winners and prize winners of the IceVision competition will share the prize pool worth RUB 3 million.
The IceVision competition was arranged as part of the series of UpGreat technology competitions in order to implement the National Technology Initiative. One of the Russian leading technological universities, NUST MISiS, acts as co-arranger, the technology partner is Skoltech. The competition targets schoolchildren, students and young professionals aged from 14 to 30 united in teams comprising 2-5 persons.
IceVision participants develop the software for installation in a driverless car able to analyze the data from stereo cameras or lidars and to correctly recognize the road situation in adverse conditions of the Russian winter: during precipitation, in counter sunlight, fog, insufficient illumination, non-cleared road. Confident recognition of signs in any weather is the prerequisite for creation of Level 5 driverless vehicles capable of moving without a driver in any weather on any public road. None of driverless vehicle manufacturers has managed to address this objective so far.
The entire list of finalist teams is available via this link.