Plenary Session
|
|
Over the decade, researchers have been investigating viable technologies to realize autonomous driving in urban environments. While large improvement has been made in technologies, there still remain many challenges that make autonomous driving hard in urban environments. Some of them are detection and tracking of a large number of moving objects in real time. Recently, for tackling these challenges, new technology based on the Deep Neural Network(DNN) has been widely adopted. While DNN has been verified as a powerful tool, it requires a significant amount of efforts in preparing data. Data pre-processing has become more important as we begin to rely heavily on sparser data sensors. In this talk, I discuss a few key issues of urban autonomous driving and introduce some recent technological advances. As well as many related core technologies, I also focus on several methods of preparing appropriate data, especially data pre-processing techniques for sparse and incomplete data. It is shown that data pre-processing is particularly useful in urban environments with many occlusions and complex road obstacles.