How AI is unlocking new levels of automotive automation
One of the ways in which artificial intelligence promises to change our lives the most is surely in the automotive sector. The promise of self-driving vehicles has ramifications in many areas, from autonomous robotaxis to driverless freight lorries. But how far off is this future?
The holy grail driving the frenzied competition between automotive manufacturers is the achievement of fully autonomous vehicles ranked level 5 on the Society of Automotive Engineers (SAE) Levels of Driving Automation Standard, meaning they are able to operate in all conditions without human interaction. To date, no solution has achieved that, with commercial offerings typically falling into levels 2 and 3. Level 2 denotes a vehicle with automated steering and acceleration features, such as stay-in-lane and self-parking, while level 3 is indicative of a vehicle capable of detecting the environment surrounding it to, for instance, overtake other vehicles.
Many, then, were surprised when Elon Musk, head of electric vehicle firm Tesla, that level 5, fully autonomous Tesla vehicles could be enabled by the end of the year - via merely a software update. addressing the World Artificial Intelligence Conference, Musk said: “I feel like we are very close. I remain confident that we will have the basic functionality for level 5 autonomy complete this year. There are no fundamental challenges remaining. There are many small problems. And then there's the challenge of solving all those small problems and putting the whole system together.”
Some remain sceptical about Musk’s claims, considering his pay is explicitly tied to Tesla’s success in the stock market. Every $50bn the company increments in value in the next ten years will see Musk receive increased remuneration, topping out at $50bn if the company is worth $650bn by 2028. Even if what he says does come to pass, there remains significant hurdles before it becomes available to the average person, with regulatory frameworks still having to be thrashed out.
Tesla undoubtedly has a habit of confounding expectations, however, and it almost every time it has been expected to fall. It is far from the only company thriving on the back of autonomous vehicle development. Google’s self driving car project was spun out into a company known as Waymo, which at the start of the year achieved of testing on public roads. The company has recently to integrate its Waymo Driver product, which can be built into third-party vehicles to afford them autonomous capabilities.
Adam Frost, Chief Automotive Officer of Waymo, said: “This key partnership with Volvo Car Group helps pave the path to the deployment of the Waymo Driver globally in years to come, and represents an important milestone in the highly competitive autonomous vehicle industry. Volvo Car Group shares our vision of creating an autonomous future where roads are safer, and transportation is more accessible and greener.”
Waymo’s partnership with Volvo is indicative of the ways traditional automotive manufacturers are fighting back after upstarts such as Tesla have as the most valuable car companies - despite selling many times less vehicles. Recognising that it is precisely technologies such as AI that are driving the future vehicle market, manufacturers like Mercedes are partnering with technology firms to introduce advanced technology to their cars.
In Mercedes’ case, it is partnering with US technology firm NVIDIA to build in-vehicle AI computing infrastructure using NVIDIA’s . The partnership will see the architecture rolled out across all next generation Mercedes-Benz vehicles, starting in 2024. Jensen Huang, founder and CEO of NVIDIA, said: “Together, we’re going to revolutionize the car ownership experience, making the vehicle software programmable and continuously upgradeable via over-the-air updates. Every future Mercedes-Benz with the NVIDIA DRIVE system will come with a team of expert AI and software engineers continuously developing, refining and enhancing the car over its lifetime.”
The other nexus of self-driving vehicle development is not in the US, but in China, where DiDi leads the way. The Chinese Uber equivalent (Having bought out Uber’s Chinese business), has received permission to test its autonomous fleet in the Jiading district of Shanghai, with vehicles on level 4 of the SAE scale, meaning they are capable of operating in all conditions without human intervention. Although disrupted by the COVID-19 pandemic, plans included a further roll out to cities such as Beijing and Shenzhen, and even California in 2021.
While there are undoubtedly challenges remaining in the path of achieving full autonomy, in terms of regulation, trust from consumers and simple development hurdles, there is only one direction in which the automotive industry is heading. From the newest challengers to the oldest, most established brands, all have realised one thing: AI is the future for vehicles.
Facebook Develops AI to Crackdown on Deepfakes
In light of the large tidal wave of increasingly believable deepfake images and videos that have been hitting the feeds of every major social media and news outlet in recent years, global organisations have started to consider the risk factor behind them. While the majority of deepfakes are created purely for amusement, their increasing sophistication is leading to a very simple question: What happens when a deepfake is produced not for amusement, but for malicious intent on a grander scale?
Yesterday, Facebook revealed that it was also concerned by that very question and that it had decided to take a stand against deepfakes. In partnership with Michigan State University, the social media giant presented “a research method of detecting and attributing deepfakes that relies on reverse engineering from a single AI-generated image to the generative model used to produce it.”
The promise is that Facebook’s method will facilitate deepfake detection and tracing in real-world settings, where the deepfake image itself is often the only information detectors have to work with.
Why Reverse Engineering?
Right now, researchers identify deepfakes through two primary methods: detection, which distinguishes between real and deepfake images, and image attribution, which identifies whether the image was generated using one of the AI’s training models. But generative photo techniques have advanced in scale and sophistication over the past few years, and the old strategies are no longer sufficient.
First, there are only so many images presented in AI training. If the deepfake was generated by an unknown, alternative model, even artificial intelligence won’t be able to spot it—at least, until now. Reverse engineering, common practice in machine learning (ML), can uncover unique patterns left by the generating model, regardless of whether it was included in the AI’s training set. This helps discover coordinated deepfake attacks or other instances in which multiple deepfakes come from the same source.
How It Works
Before we could use deep learning to generate images, criminals and other ill-intentioned actors had a limited amount of options. Cameras only had so many tools at their disposal, and most researchers could easily identify certain makes and models. But deep learning has ushered in an age of endless options, and as a result, it’s grown increasingly difficult to identify deepfakes.
To counteract this, Facebook ran deepfakes through a fingerprint estimation network (FEN) to estimate some of their details. Fingerprints are essentially patterns left on an image due to manufacturing imperfections, and they help identify where the image came from. By evaluating the fingerprint magnitude, repetition frequency, and symmetrical frequency, Facebook then applied those constraints to predict the model’s hyperparameters.
What are hyperparameters? If you imagine a generative model as a car, hyperparameters are similar to the engine components: certain properties that distinguish your fancy automobile from others on the market. ‘Our reverse engineering technique is somewhat like recognising [the engine] components of a car based on how it sounds’, Facebook explained, ‘even if this is a new car we’ve never heard of before’.
What Did They Find?
‘On standard benchmarks, we get state-of-the-art results’, said Facebook research lead Tal Hassner. Facebook added that the fingerprint estimation network (FEN) method can be used for not only model parsing, but detection and image attribution. While this research is the first of its kind, making it difficult to assess the results, the future looks promising.
Facebook’s AI will introduce model parsing for real-world applications, increasing our understanding of deepfake detection. As cybersecurity attacks proliferate, and generative AI falls into the hands of those who would do us harm, this method could help the ‘good guys’ stay one step ahead. As Hassner explained: ‘This is a cat-and-mouse game, and it continues to be a cat-and-mouse game’.