Principal Investigator Dina Katabi
Co-investigator Daniela Rus
Project Website http://toyota.csail.mit.edu.ezproxy.canberra.edu.au/node/39
Automatic detection of people, cars, and obstacles is critical for future driver assistance systems and robot navigation. This is particularly important when the object is not in the field-of-view of the car and its driver. Many accidents happen today due to inability to detect a child running around the corner or a car pulling out of an occluded driveway, as shown in the figure below. The future of driver assistance systems and autonomous cars critically depends on how well they can deal with blind spots and hidden objects.
Unfortunately, existing sensing technologies based on vision and LIDAR systems cannot operate in non-line of sight, and hence fail to deal with such scenarios. This motivates the need for additional sensing technologies that can operate through occlusions and around corners. Thus, the purpose of the proposed research is to develop a system that enables detecting cars and people through occlusions.
To address this problem, we propose to use WiFi signals as a sensing modality of people and cars. Wireless signals can traverse occlusions and operate in non-line-of sight. Hence, if one can locate and track cars and people using wireless signals, one can avoid running into a car that emerges from a hidden drive way or a child that suddenly appears around the corner.
We propose a comprehensive research project where a WiFi-based sensing system in the car detects and tracks other cars and people in the scene. The research considers both the scenario in which the surrounding cars and people have WiFi devices on them, and the scenario in which they lack any source of WiFi signals. It also considers the presence and absence of roadside sensors. We will also develop algorithms that realize safe motion using such feedback.
Furthermore, we consider how such sensors can help in detecting the speed of an incoming car, the presence of an empty parking spot, and other important traffic metrics. The proposed solutions will also be applicable more broadly to applications requiring indoor localization, and hence be used for robot navigation in buildings and factories.