Driving SMARTS Competition
Driving SMARTS Competition aims to heighten research attention from the machine learning (ML) community to challenges resulting from distribution shift in diverse dynamic interaction contexts prevalent in real-world autonomous driving (AD). This competition is designed to support methodologically diverse solutions, such as reinforcement learning and offline learning methods, using a combination of naturalistic AD data and open-source simulation platform SMARTS.
The two-track structure allows focusing on different aspects of the distribution shift. Track 1 is open any method and will give ML researchers with different backgrounds an oppotunity to solve a real-world AD challenge. Track 2 focuses strictly on offline learning methods. The emphasis on real-world scenarios and multi-faceted evaluation metrics promote development of more realistic and human-friendly behaviors, allowing, as a result, a principled investigation into the opportunities and dangers in large scale deployment of autonomous.
© Copyright Huawei. All Rights Reserved