Imagine instructing your car to drop you off at a specific location. Imagine you also tell it to find a parking space. Think of the time and mental energy you would save. This scenario, envisioned by Dr. Carl Crane, a professor in the UF Department of Mechanical & Aerospace Engineering, may become a reality in the not too distant future.
Research on automated technology has been active for more than 20 years at the Center for Intelligent Machines & Robotics (CIMAR), which is headed by Crane. Two major events that led to important advances in this technology in his lab included competing in the U.S. Defense Advanced Research Projects Agency (DARPA) Grand Challenge in 2004 and 2005 and then later in the 2007 DARPA Urban Challenge.
“We developed a method for an autonomous vehicle to decide on what is the best behavior to do at any given instant,” Crane said. “We call our approach the Adaptive Planning Framework, and we developed it and tested it extensively during the DARPA Urban Challenge.”
Connected and automated vehicles are on the U.S. Department of Transportation (USDOT) research agenda, and already five states (California, Florida, Michigan, Nevada and Virginia) have been approved for on-road testing. Autonomous vehicles do currently exist, and are being tested, but they do not necessarily have the ability to communicate with the infrastructure or with other vehicles in their vicinity. Connectivity and infrastructure enhancements can increase the benefits of vehicle automation. Proponents of these technologies believe roads will be safer and congestion will be alleviated, thus reducing commute times.
Crane, a member of the UFTI’s Internal Steering Committee, is co-PI on a project funded by the Florida Department of Transportation (FDOT) (PI is Dr. Lily Elefteriadou, director of the UFTI). It is expected this project and others at UFTI, for that matter, will carry this vehicle technology one step closer to becoming commonplace on our roadways.
“Currently, we are working on a project where autonomous and human driven vehicles will be sharing the road,” Crane said. “The autonomous vehicle will communicate its speed, position, and intent (go straight, turn left, or turn right) with the traffic light controller. The controller will use this information to improve traffic flow.”
Crane and his team at CIMAR recently completed a project which involved autonomous vehicles and standard cars driven by human subjects. In this study, the autonomous vehicles had to predict or make a decision on how to proceed based on the behaviors of the human-driven vehicles approaching it.
“For example, at an intersection, a vehicle’s intent would be whether it wants to go straight or turn left or right,” Crane said. “A vehicle’s belief would be its perceived model of the environment. For example, has the other vehicle seen the autonomous vehicle or not and added it to its model of the environment?”
Crane says the autonomous vehicle will drive with more caution, if it assesses or estimates the other vehicle’s belief may be incorrect.
“Tests in simulations have shown that this approach of estimating intent and belief have resulted in a reduction in collisions,” he said.
Consistent with the USDOT’s goal that advanced technologies must be safe and reduce accidents, Crane foresees the UFTI playing a major role in creating a relatively benign technology to implement on city roads and highways.
“Safety is always the number one problem,” Crane said. “An autonomous vehicle that makes correct decisions 99% of the time is not good enough. Highway designs will have to account for what autonomous vehicles can do well and avoid scenarios where sensors might fail or give false readings. A question is how to best do this without adding unnecessary costs to the infrastructure.”
In due course, automated and connected vehicles will be traveling safely and efficiently along urban environments, making commutes shorter and improving many aspects of life as we know it. And what’s more, UFTI affiliated researchers will be at the forefront of this emerging transportation technology.
So, there will be a day when driving around trying to find a parking space to get to work or to an appointment will be a thing of the past.
“That would be great to have here at UF,” Crane said.
Carl D. Crane is a Professor in the Department of Mechanical and Aerospace Engineering and Director of the Center for Intelligent Machines and Robotics (CIMAR) at the University of Florida. He has been at UF since January 1987. Dr. Crane has been involved in research in the areas of spatial mechanisms, tensegrity systems, robotics, and autonomous navigation for over twenty five years. Current activities include the development and implementation of system architectures for autonomous ground vehicle navigation and the design and implementation of passive parallel mechanisms to be used for force control applications. Dr. Crane was team leader of the University of Florida’s 2004 and 2005 DARPA Grand Challenge autonomous vehicle development efforts and for the University of Florida’s 2007 DARPA Urban Challenge project team. Dr. Crane has authored one book and over forty papers in the area of spatial geometry and robotic systems. He has supervised one hundred and ten Master’s graduates and thirty three Ph.D. graduates during his career at the University of Florida.