Sensor technologies used in self-driving vehicles - such as cameras, radar, sonar and LiDAR (light detection and ranging) - are all vulnerable to hacker strikes, say the firms. The companies, working with the University of Virginia School of Engineering (UVa), have run tests that they say demonstrate how it is relatively easy to hack into such systems and disrupt critical functions such as braking and accelerating.
A test was conducted earlier this year on a "reference architecture" autonomous ground vehicle supplied by Perrone Robotics. Originally designed for participation in the DARPA Urban Challenge, the vehicle - a modified Scion xB - self-navigates streets, plans and follows routes, operates at intersections, avoids collisions with vehicles, self parks, and implements follower and collision avoidance modes of operation.
During the test (see video below), which was part of a pilot project testing the capability of MSi's Secure Sentinel platform, MSi and UVa personnel were able to hack into the unprotected vehicle's speed control functions using a key fob attack, causing the vehicle to collide with a soft target. The test was also able to demonstrate, say the companies, that it is relatively easy and inexpensive to protect against such attacks.
In this case, MSi's Secure Sentinel platform - which consists of embedded hardware with the physical system and cloud-based software - was shown to be able to detect and foil such attacks. "We successfully demonstrated [that], yes, you can easily attack braking and acceleration and other automated features added to cars, but you can also protect against them in real time," said David Drescher, CEO of Mission Secure.
The product is being made available to the U.S. military, energy and transportation industries. For the automotive market, the goal is to reduce the cost to $15 and potentially evolve it to a smartphone application. Installation on new vehicles could take place after extensive testing and regulatory requirements are met. For more,