Take a minute to imagine that all vehicles on the planet are autonomous. You jump in your car, shout out your desired destination (although I would probably have to type it due to my thick Scottish accent), sit back, and enjoy the ride. How long do you think it will it be before the last human-driven car is allowed on public roads? 50 years? 100?
“Distinguished members of IEEE, the world’s largest professional organization dedicated to advancing technology for humanity, have selected autonomous vehicles as the most promising form of intelligent transportation, anticipating that they will account for up to 75 percent of cars on the road by the year 2040.”[1].
There is no denying autonomous vehicles would bring a lot of benefits. You’d be free to spend your time in the car doing whatever you like: catching up on emails, reading, sipping a pint, sleeping. Also, I got to thinking how much safer I’d feel travelling back from a night out on the town in a car operated by a rigorously engineered and tested computer, as opposed to a total stranger. I guess taxi drivers would go extinct?
It would revolutionise the lives of elderly and disabled people otherwise unable to get from A to B without relying on a driver.
Then there is of course the inevitable massive reduction in road accidents by no longer having us easily-distracted humans behind the wheel. According to the World Health Organisation, ‘Road traffic injuries are the leading cause of death among young people, aged 15–29 years. About 1.25 million people die each year as a result of road traffic crashes’ [2].
It is true to say that most road accidents are due to human error. If you remove the human, there is far less chance of error and therefore accidents. However, this argument relies heavily on all vehicles that share the road being reliable, resilient and secure. As we have learned over the years, software bugs can be disastrous (for example see software bug disasters). Also, as soon as an autonomous vehicle is compromised and in an attacker’s hands, this argument flips on its side: and it could be said an attacker can do more harm the less a human is able to intervene.
After considering all the benefits I got to thinking about what we’d lose in an autonomous world of travel.
Imagine never again being able to jump on one of these to ‘just drive’.
Would we ever be able to get the same sense of freedom in such a safe and unspontaneous world of autonomous travel?
The google self-driving car prototype has no steering wheel, accelerator or brake pedal.
In a society full of CCTV, email snooping, phone tapping, tracking, data mining and profiling, it can be unclear how much of your personal life is actually personal. I find it liberating to know that I have the option of driving into the middle of nowhere and going off the radar for a bit. The thought of this option being taken away makes me feel claustrophobic.
Cars have been moving towards full autonomy for quite some time. New embedded systems and features get added with each new release. You can currently buy cars with crash avoidance, lane keeping, self-parking, smart cruise control, in-car navigation, vocal texting, in-car wifi, and you can remotely control features such as the engine and the locks.
The more technologically advanced a vehicle becomes the more targeted attacks they will inevitably face. One proof of concept attack which gained a lot of press in 2015 was when FIAT Chrysler had to recall 1.4 million cars after US-based security testers Charlie Miller and Chris Valasek proved it was possible to take control of a Jeep Cherokee remotely via its internet-connected entertainment system, and control the steering, brakes, dashboard functions, and transmission,all from a laptop miles away (see their talk at Black Hat USA 2015 - Remote Exploitation Of An Unaltered Passenger Vehicle). Many companies believed, until then, that they were safe against such attacks because they’d ensured separation of the internet-enabled entertainment area and the Controller Area Network (CAN). Miller and Valasek bypassed this separation. Their research showed that whilst it wasn’t possible to simply connect to the car’s entertainment system and send messages to the CAN, it was possible to perform a firmware update on the chip which can talk to the CAN so that it contained backdoored firmware which accepted data passed in through the entertainment system and would send messages to CAN. Hence allowing for control of the car, all remotely from a laptop connected to the same cellular network, Sprint (which covers most of the US). The cellular network Sprint and Chrysler both took action to prevent this attack.
Miller and Valasek were fair to highlight the fact that this wasn’t a Chrysler-specific issue, and it was just that their research focussed on the one company. They have since done further research covering more manufacturers and released a ‘How hackable is your car’ chart (see http://www.wired.com/2014/08/car-hacking-chart/).
Another attack which made the headlines was when car thieves were able to access a diagnostic port in BMWs to reprogram blank keys and gain possession of keys to expensive BMWs; all in a matter of minutes (http://www.bbc.co.uk/news/uk-19562487).
Most modern cars have WIFI, 3 or 4G, Bluetooth, and sometimes even USB connections. It is essential that in an industry so target-driven and competitive, amongst the excitement of adding new gadgets, that security be at the forefront of the design, and not be an expensive after-thought. Research such as that performed by Miller and Valasek helps bring security to the full attention of car manufacturers. Following these attacks, a US senator wrote a letter to a number of car manufacturers prompting them to answer questions about their security testing process. The letter stated:
‘As vehicles become more integrated with wireless technology, there are more avenues through which a hacker could introduce malicious code and more avenues through which a driver's basic right to privacy could be compromised.
These threats demonstrate the need for robust vehicle security policies to ensure the safety and privacy of our nation's drivers.’
On behalf of the car manufacturers, The Auto Alliance responded with:
‘Auto engineers are incorporating security solutions into vehicles from the first stages of design and production, and their security testing never stops.
As cars and other forms of transportation increasingly incorporate in-vehicle computer systems to help with everything from safety to navigation, cyber-security is among the industry’s top priorities and the auto industry is working continuously to enhance vehicle security features.’
Tesla expect to have developed a fully self-driving car in just a few years’ time [3] and has taken a direct and proactive approach to security by appearing at one of the world’s largest hacker conference in 2015 with a Model S and an offer of $10,000 as a bug bounty. They were also enthusiastic in trying to recruit from the hacking community. Personally I think this is a wise move which all car manufacturers should consider. New software needs to be continually hardened against new threats, and finding and addressing vulnerabilities before malicious attackers do is something that the hacking community could help with, if some rapport is built and incentives are offered. Some are a little behind on this front, with no mention on their respective web pages explaining how hackers can report security vulnerabilities.
Security-wise there are, as always, two main concerns: is the vehicle secured against compromise? What data could be stolen? The recommended security defences do not differ to that of any other software environment: implement security into the software development lifecycle and give it a lot of consideration at the design stage, have the environment security tested both internally and by a third party continuously throughout development, segregate as much as is possible, install security software updates immediately. However, it is the method by which some of these defences are implemented which will differ to traditional networks. For example, vehicles need to be able to be updated quickly and easily, remotely. Physical updates won’t do.
In a world of autonomous travel, it is highly likely that you would be tracked wherever you go. It is important that we determine suitable laws which protect privacy from being extinguished on the roads too. If tracking data got into the wrong hands, an attacker would potentially be able to identify the user’s home address, current location, favourite hangouts… The user’s past travel patterns and potentially his or her future travel plans. It could be determined that you visited a strip club on the Friday, a fine restaurant on Saturday, your local church on Sunday and back to work at the houses of parliament on the Monday. All legal, but an example of the types of information that most would want to remain private. The imperative of balancing consumer privacy with the advancement of autonomous vehicle technology becomes an urgent one.
References:
[1] http://www.ieee.org/about/news/2012/5september_2_2012.html
[2] http://www.who.int/mediacentre/factsheets/fs358/en/
Other sources
http://www.fastcompany.com/3024362/innovation-agents/10-autonomous-driving-companies-to-watch
http://www.bbc.com/future/story/20141013-convoys-of-huge-zombie-trucks
http://www.ieee.org/about/news/2012/5september_2_2012.html
http://www.wired.com/2015/10/five-car-hacking-lessons-we-learned-this-summer/
http://www.ni.com/white-paper/2732/en/
https://www.youtube.com/watch?v=KhqL2QHSkeM
Categories