Ben-Gurion University of the Negev is one of Israel’s leading research universities and among the world leaders in many fields. It has around 20,000 students and 4,000 faculty members. The Faculty has three main campuses: The Marcus Family Campus in Beer-Sheva; the research campus at Sde Boker and the Eilat Campus, and is home to national and multi-disciplinary research institutes: the National Institute for Biotechnology in the Negev; the National Institute of Solar Energy; the Ilse Katz Institute for Nanoscale Science and Technology; the Jacob Blaustein Institutes for Desert Research; the Ben-Gurion Research Institute for the Study of Israel & Zionism, and Heksherim – The Research Institute for Jewish and Israeli Literature and Culture.
Camera sensors are emerging in many applications such as Smart City and Autonomous Driving. The Data generated by multiple cameras in a smart city and autonomous driving applications is usually transmitted through an edge box to a cloud and viewed in a control room terminal. This transmitted information requires a considerable high channel bandwidth, which is not always available using standard current communication means. In this research we are proposing to explore innovative new algorithms to take advantage of video content in order to reduce bandwidth requirements for a given video quality. In addition we are proposing ways to compensate for non-ideal quality by providing the remote driver of autonomous vehicles assistance in rapidly assessing the situational awareness picture surrounding the vehicle and responding faster to potential road hazards. The primary method that is proposed for reducing bandwidth requirement is based on detecting a Region of Interest (ROI), to which the Human Visual System (HVS) is drawn. Such ROIs can then be used to selectively adjust Rate Control parameters and improve bandwidth utilization. The detection of ROIs in images is a researched area. However, performance is becoming a major constraint when processing video streams and having to make these decisions in real time. We are proposing to substantially improve ROI detection performance. In addition, we will use detected objects’ coordinates to indicate to remote drivers the most important objects in the view, thus allowing them to reach fast navigation and steering decisions despite occasional non-ideal video resolution. The same objects’ coordinates will also be used for subsequent AI algorithms in order to improve their performance and accuracy results.
The primary purpose of the BGU research is to contribute to the consortium an efficient means of delivering live streaming video from multiple platform cameras in real time. This video will allow the operation to overcome unstable channel bandwidth allocation and support remote platform navigation despite the challenges that are introduced by such channels. The proposed research project will focus on three different directions as follows:
These research goals fit well in the consortium’s overall mission to create an echo system for remote supervision, steering and navigation of autonomous platforms (whether they are vehicles, drones or others).