Overview

Self-driving will potentially enhance vehicle safety and performance as well as traffic efficiency. One of the key objectives of self-driving vehicles is to achieve safe and robust decision making under dynamic and complex environments, especially in dense traffic environments. To achieve this objective, self-driving vehicles firstly need to understand scenes in the environments and then use this scene understanding to make decisions and perform path planning. Furthermore, self-driving vehicles need to have learning capabilities to learn from past experience and reuse the learned knowledge to continually improve driving performance so that they can handle uncertainties and dynamics from moving objects, road and weather conditions in real-world traffic, especially unknown traffic conditions. In addition, learning-based systems may make decisions that are not easily explainable to human users, and their safety is difficult to verify. Also, the learning capacity is hard to validated and verified due to the outlier events. This workshop seeks to explore areas related to these challenges.

Organizers

Jiucai Zhang (GAC R&D Center in Silicon Valley Inc., USA)

Xiaoli Zhang (Colorado School of Mines, USA)

John Dolan (Carnegie Mellon University)

Xiao Lin (GAC R&D Center in Silicon Valley Inc., USA)