Huawei's Mate 10 Pro recently became the world's first phone to drive a car but that demo should only be seen as "hype" and isn't indicative of a commercially feasible solution being on the horizon, Avast Research Director Martin Balek told AndroidHeadlines. While the experiment that saw the Chinese tech giant modify a Porsche Panamera in order to have it accept input from its Android phablet was already presented as a marketing stunt and not a product announcement, Mr. Balek stressed the importance of having low expectations in regards to the very concept of a smartphone capable of driving a vehicle. Using a standard commercial handset — high-end or not — as the center of a self-driving solution wouldn't be "wise" even if the device itself may have enough power to actually process the artificial intelligence programmed to navigate the roads, according to the industry veteran.
Adequate security is effectively an insurmountable obstacle to the commercial viability of such a technology as no single smartphone or any other device can replace the multi-layer protection mechanisms required by standards utilized in something like the aviation industry, many of which should also be adopted by the self-driving segment, according to the cybersecurity expert. Things like fail recovery and redundancy solutions require multiple hardware units powering independent systems capable of checking their readings and performance results against each other in order to ensure everything is working as expected, so attempting to replace such setups with a single device would likely endanger the overall security and reliability of self-driving vehicles, regardless of whether that module is an ultra-premium smartphone, a console, or something else entirely, Avast's official indicated.
Hypothetical plug-and-drive applications such as the one demonstrated by Huawei and Porsche last month would likely be more vulnerable to man-in-the-middle attacks, i.e. vectors that don't target neural processing units like the one found inside the Mate 10 Pro that's controlling the car but seek to exploit the communications channels between the phone and vehicle in order to seize control of the latter, Mr. Balek explained. Even if an attacker was to target the smartphone itself, AI chips aren't any safer than regular ones, though that isn't as big of an issue as it may sound, according to the same expert.