Professor Brian C. Williams
PI at MIT
Professor Majid Khonji
PI at MIT
For the past hundred years, innovation within the automotive industry has created more efficient, affordable, and safer vehicles, but progress has been incremental so far. The industry now is on the verge of a substantial change, caused by Autonomous Vehicle (AV) technologies. This technology offers the possibility of significant benefits to the society, including saving lives and reducing crashes, congestion, and pollution. A significant barrier to deploying AVs on a massive scale is safety assurance. Several technical challenges arise from the uncertain environment in which AVs operate, such as road and weather conditions, as well as errors in perception and sensory input data. A robust AV control algorithm should account for different sources of uncertainty and should generate control policies that are quantifiably safe. Algorithms that respect precise safety measures can also assist policymakers to address legislative issues related to AVs, such as insurance policies and road network usage along with non-autonomous vehicles. A single vehicle always has a limited scope of perception or potential blind spots. Vehicle-to-vehicle (V2V) and Vehicle-to-Infrastructure (V2I) are technologies that enable vehicles to exchange safety and mobility information with each other and with the surrounding infrastructure, such as traffic lights. V2V messages would have a range of approximately 300 meters, which exceeds the capabilities of systems with cameras, ultrasonic sensors, and LIDAR, allowing greater capability and time to warn vehicles. In addition, these radio messages can “see" around corners or "through" other vehicles, addressing scenarios such as those, for example, where an oncoming vehicle emerges from behind a truck or from a blind alley. In those situations, V2V communications can detect the threat much sooner than radar or camera sensors can. In this project, each AV has to be able to plan trajectories from its current location to goals while avoiding static and dynamic (moving) obstacles, while meeting deadlines and efficiency constraints. The risk of collision should be bounded by a given safety threshold that meets governmental regulations, while meeting deadlines should achieve a quality of service threshold. Besides, we consider optimizing fleet-level operation by coordinating actions among vehicles equipped with long-range unicast communication (4G/Beyond 4G). Our planning algorithms should account for potential communication delays and link uncertainty. .
The techniques studied in this work will complement future research projects in AV (as well as in
marine applications) at Khalifa University Center for Autonomous Robotic Systems (KUCARS). A
future showcase of this work can extend to an actual deployment of our algorithms on a small fleet of
autonomous vehicles in a controlled environment. The project is in line with UAE Artificial Intelligent
strategy and Dubai's Autonomous Transportation Strategy 2030. The algorithms developed in this
work will be tested using a simulation testbed that relies on UAE road-network topology and road
conditions.
