This talk will identify numerous challenges to make self-driving service a reality, including data annotation at scale. By applying GPU-based AI and ML to automatically annotate data accuracy can be increased. In addition, multi-modality-based perception can be utilized to improve confidence. Driving policy in complex environment is another intriguing challenge which motivates development in the field of reinforcement learning (RL). Specific aspects of RL will be discussed, such as the need to act in very high dimensional spaces and to make decisions with long-term consequences.
Hall: Hall D