Autonomous execution of the Precision Immobilization Technique

- Autonomous vehicles present a very unique cyber-physical security challenge.
- A control law for an autonomous car to execute the PIT maneuver was developed.
- The PIT maneuver control law was demonstrated in simulation as well as experimentally.

Over the course of the last decade great advances have been made in autonomously driving cars. The technology has advanced to the point that driverless car technology is currently being tested on publicly accessed roadways. The introduction of these technologies onto publicly accessed roadways not only raises questions of safety, but also security. Autonomously driving cars are inherently cyber-physical systems and as such will have novel security vulnerabilities that couple both the cyber aspects of the vehicle including the on-board computing and any network data it makes use of, with the physical nature of the vehicle including its sensors, actuators, and the vehicle chassis. Widespread implementation of driverless car technology will require that both the cyber, as well as physical security concerns surrounding these vehicles are addressed. In this work, we specifically developed a control policy to autonomously execute the Precision Immobilization Technique, a.k.a. the PIT maneuver. The PIT maneuver was originally developed by law enforcement to end high-speed vehicular pursuits in a quasi-safe manner. However, there is still a risk of damage/roll-over to both the vehicle executing the PIT maneuver as well as to the vehicle subject to the PIT maneuver. In law enforcement applications, it would be preferable to execute the PIT maneuver using an autonomous vehicle, thus removing the danger to law-enforcement officers. Furthermore, it is entirely possible that unscrupulous individuals could inject code into an autonomously-driving car to use the PIT maneuver to immobilize other vehicles while maintaining anonymity. For these reasons it is useful to know how the PIT maneuver can be implemented on an autonomous car. In this work a simple control policy based on velocity pursuit was developed to autonomously execute the PIT maneuver using only a vision and range measurements that are both commonly collected by contemporary driverless cars. The ability of this control policy to execute the PIT maneuver was demonstrated both in simulation and experimentally. The results of this work can help inform the design of autonomous car with regards to ensuring their cyber-physical security.