Automobile manufacturers are making use of artificial intelligence (AI) to improve the problem-solving capabilities of modern-day vehicles.
Newer automotive features like adaptive cruise control, traffic sign recognition and automatic electronic break systems all use AI. AI allows automotive systems to adapt to new circumstances that were not forecasted during the hardware’s design phase. It also has the inherent ability to come up with a solution to a problem much faster than alternative implementations, while still guaranteeing the robustness and the integrity of the solution.
These capabilities make AI a cornerstone to bringing autonomous driving into the mainstream, but let’s discuss how frequent system updates as well as robust system-level security are required to take full advantage of AI in automotive systems.
Frequent Updates Are Essential
Autonomous cars need constant updates — not all of them related to AI. For example, a vehicle owner may purchase new services or features that require updates. Or malfunctioning equipment might require a local (in-car) update by the auto dealership using the over-the-air/secure over-the-air (OTA/SOTA) methodology.
When an automotive system uses AI, it requires very frequent updates. If the system has self-learning capabilities, for instance, an extended period on the road may require a new set of synaptic weights and biases* to further improve autonomous driving capabilities. The new combination of inputs and outputs must be sent to the factory server to generate the new AI parameters. The factory then uses OTA/SOTA to update all autonomous vehicles deployed in the field so they can start benefitting from the new parameters.
Regardless of the type of update, all updates to automotive systems must be secure and originate from a certified authority.
Security Must Be Tight
In autonomous driving, safety is the number one priority. And because implementing AI in automotive systems requires the manipulation of large amounts of data that could lead to a change in the behavior of the vehicle — and in extreme cases, threaten the safety of the driver, the passengers and, potentially, the people and property around vehicles — the data must be protected from unauthorized (whether malicious or accidental) changes.
This data is stored in a nonvolatile memory device that must be constantly written, read and updated in a secure way that’s not easy to breach. That’s why Micron is investing a lot of time, research and development into memory solutions that enable the ultra-secure transfer and storage of data.
How Micron Can Assist
With over 25 years in the automotive industry, Micron has been working alongside our partners and customers to provide memory and storage solutions to meet some of the industry’s toughest challenges. Check out my next post where I’ll take a walk down memory lane to look back at the attempts over the last few decades to protect data from harm. Then, I’ll talk about the more recent protection schemes available, including Micron’s Authenta™ technology, aimed at protecting against automotive system attacks.
Discover more about Micron's Automotive Solutions now.
*Artificial Intelligence is depending from the weight matrixes and the initial conditions. The bias is a precondition value of the neurons (if any) of the Neural Network behind the Artificial Intelligence.