Description

It a tech talk on the topic Multi-Modal Sensor Data Processing for autonomous vehicles organized by IEEE CIS student branch chapter of UOJ. In today's talk, Prof. Thia Kirubarajan will consider the problem of using multimodal sensor data with application to advanced driver assist systems (ADAS) and autonomous vehicles (AV). Improving the safety and security of ADAS and AVs requires the processing of different types of data (e.g., radar, video, acoustic, lidar) to detect obstacles in a timely manner. Because of the need to ensure safe operation under all weather and lighting conditions, it is critical to develop and implement algorithms that can work with multimodal data. The obstacle detection module in ADAS or AV systems consists of numerous modules, including data preprocessing, object detection, classification, prediction, tracking, and fusion. He will review the challenges in the data processing modules of ADAS and AV systems and discuss different algorithmic choices for these modules and present some results.