AI development is accelerating. In 2016, AlphaGo, developed by Google DeepMind, beat one of the world's leading Go players. AlphaGo has independently played and analysed several million rounds of Go, in order to learn the game and introduce strategies that no human beings have previously formulated. Go is a complicated Asian board game.
In June, Google demonstrated the world's so far largest artificial neural network, with more than one billion neurons. This network is used to identify images, and was 70% more successful than previous attempts. When the network described images based on a thousand different categories, it was accurate in 50% of cases.
AI is increasingly being used within healthcare. This gives a better understanding of learning and of how robots can be designed to feel like human beings. This has led to surgical procedures that are far less invasive than before.
AI has also come far within the automotive industry. Self-driving cars have long been Utopia, but are now approaching achievement. Car models with autonomous driving functions already exist today and during the next few years many car models with autonomous driving functions are expected to be presented by several car manufacturers.
There is a long way to go to fully autonomous vehicles that can travel to the final destination without any driver involvement at all. Most of the technology already exists, but greater reliability still needs to be developed, besides the creation of standards and adjustment of legislation to allow for fully autonomous vehicles.
The first step is semi-autonomous driving whereby the car handles certain functions, while the driver participates actively and holds the overall responsibility. Over time, the degree of autonomy is expected to increase, ending up with fully autonomous vehicles.
The development trend is from passive to active traffic safety. The automotive industry is developing from protecting drivers and passengers, to preventing accidents. Today, 90% of all accidents are related to the human factor, and therefore active safety solutions are being developed, such as advanced vehicle assistance systems, vehicles with a high element of automation, and gradually also self-driving cars. Coming generations will use both online- and offline-based machine learning whereby information is drawn from a number of data sources in order to develop the relation with the individual driver.
At the CES electronics fair in Las Vegas in January 2017, Autoliv for the first time presented its research platform with technologies that Autoliv can already offer customers today. LIV, or the Learning Intelligent Vehicle, is based on a Volvo XC90, with equipment that includes eyetracking technology from Smart Eye.
In Sweden, Drive Me is being prepared as the world's first large-scale trial in which self-driven cars will run on ordinary roads. This project was initiated by Volvo Cars together with the Swedish Transport Administration, the Swedish Transport Agency, Lindholmen Science Park and the City of Gothenburg. The project concerns 100 vehicles, and initially 30 of them will be equipped with technology from e.g. Smart Eye.
Together with NVIDIA, Smart Eye is delivering a camera system for the vehicles. The system is cost-effective and the cameras are so small that they can easily be seamlessly integrated into the instrument panel.
The self-driven cars in the Drive Me project will be equipped with logger systems, and with the new type of cameras, as well as more traditional camera designs. The project concerns a platform which allows for recording, compression and advanced in-car AI functions such as eyetracking.
The rapid development of new technology makes companies like Smart Eye well-prepared to contribute to reducing the number of traffic fatalities and injuries.