Washington DC - Hackers pose a real danger to self-driving vehicles, US experts are warning, and carmakers and insurers are starting to factor in the risk.
Expected on the road by 2020 or even sooner, driverless cars will have a wide range of cutting-edge technologies such as electronic sensors - a group of cameras, radar, sonar, and light detection and ranging (Lidar) - commanded remotely using software that senses road widths, identifies signs and even roadblocks.
But like connected vehicles and their onboard multimedia systems, these new self-driving technology elements - which were meant to make the cars safe and reliable - could end up leaving them vulnerable to hacker strikes, according to US security firms Mission Secure and Perrone Robotics.
A hacker recently boasted of having entered the electronic systems of the US jet he was traveling on, and of having changed its trajectory. He claimed he did so using the in-flight Wi-Fi system.
The two security companies, working with the University of Virginia and the Pentagon, have run tests that have shown they believe it is possible to hack into and disrupt the multi-sensor system.
One trial was to change how the car responded when it encountered an obstacle.
According to a report available on MSi's website: ”One attack scenario forces the car to accelerate, rather than brake, even though the obstacle avoidance system (using Lidar) detects an object in front of the car. Rather than slowing down, the car hits the object at high speed, causing damage to the car and potential threat to the life and safety of the passengers in the car under attack and in the car being struck.
“If an attack were carried out successfully, manufacturers have no means of quickly gathering information for forensic analysis or to rapidly deploy additional protections to cars in response to new and evolving attacks,” the report warns.
According to these experts, hackers penetrate the system through wireless connections.
MSi and Perrone Robotics, which are working on a system to counter cyber attacks, believe the situation poses “significant challenges and risks for the automotive industry, as well as to public safety”.
Most of the carmakers gearing up their own autonomous car projects did not reply to update requests from AFP.
But sources close to the industry say the chance of the system being hacked has been factored in throughout the manufacturing process.
HACKING ITS OWN SOFTWARE
Google, for example, is believed to have a team of top programmers tasked with trying to hack into the software in its own self-driving prototype car, which is expected to begin on-road testing within the next few months. Google declined to comment for this story.
US insurers are concerned about safety, and whether the new technologies can cut the risk of accidents happening, which could force them to rethink their contracts and to recalculate premiums.
Insurer Nationwide said premiums could ride initially, because the price of self-driving cars will be high due to the cost of embedded technologies and repairs. However, this could be partially offset by the wider use of cars decked out with accident-preventing technologies.
For State Farm, another US insurer, the big picture is what counts.
“As connected and automated vehicle technology reduces or eliminates some risks that drivers face today, new risks are likely to emerge. We are focused on the big picture - how can we adapt to these changes and continue to deliver value to our customers,” the company said.
According to a source with a major US insurer, who requested anonymity, one of the key related issues down the road will be establishing boundaries, and responsiblities based on what carmakers say the car can or cannot do autonomously.
Google announced in May that its self-driving prototype cars were ready to leave the test track and hit public roads in California, in a big step forward for its autonomous automobile program.