The basic objective of its application is to improve road safety and efficiency.

  • Simple artificial intelligence is already present in many of the cars we drive, especially in some ADAS safety systems and digital assistants.
  • Cameras in ADAS systems are installed in the windshield and need to be recalibrated after a replacement.
  • AI with machine learning and deep learning is a key element in the development of the autonomous car.

Although it still sounds like science fiction, artificial intelligence (AI) is already present in many of the cars we drive. And its integration in cars will grow in the coming years, as it is the key element of safety systems, future autonomous driving and many mobility-related services.

There are currently cars on the market with simple artificial intelligence systems (without learning capabilities), used in digital assistants and in some functions of ADAS safety systems. The latter already offer artificial vision through the camera mounted on the windshield, other sensors and image processing algorithms. Thanks to this, they can recognize the environment, identify risk situations and detect, for example, road markings, signs, pedestrians or cyclists. Carglass reminds you that after replacing the windshield of a car equipped with ADAS safety systems, the camera must be recalibrated to provide correct information to the system.

On the other hand, the digital assistants incorporated in some car models use artificial intelligence for natural language processing so that we can communicate with a machine using the way we speak. In this way, instead of telling the car “raise the temperature of the air conditioning to 24 degrees”, we can say “I am cold”, so that it can perform this operation. In addition, these systems recognize routines (usual routes, favorite music, favorite temperature…) to automate them.

AI with learning capabilities

The next steps for artificial intelligence in cars is to have the capacity to improve through machine learning and deep learning, which are vital for the autonomous car.

Hyundai has already developed the world’s first AI-based ADAS function with machine learning. This is a cruise control that recognizes, analyzes and learns from the driver’s driving patterns to maintain distance from the car in front, accelerate and respond in a way that is identical to what the vehicle owner would do. In this way, the driver feels that the car reacts as he himself would, and does not feel reluctant or uncomfortable using this system.

AI with deep learning is not yet on the market, but it is in the development of the autonomous car being carried out by different players. Wayve, for example, managed to get a car to learn to drive without leaving the road in just 20 minutes, after twelve attempts and numerous corrections by the human driver. According to this British startup, they have developed the first autonomous car capable of driving in real traffic using only cameras and sensors, AI and a GPS navigator. Their autonomous car uses deep learning to learn to drive by experience, example and feedback, just like a human being does. They are already conducting tests in real traffic to perfect this technology.

But the development of the autonomous car goes much further, as a self-guided car has to define an environment and context, based on the information it has gathered, in order to make the right decisions. Bosch explains that this requires it to learn from experience and to predict how a situation will develop. To give a simple example, when it detects a ball rolling between two parked cars, it has to know that a child might run after it and slow down as a precaution.

Engineers teach the car’s AI the mathematics, physical laws, biology (body shapes of people and animals) that it will have to apply in its daily life. And the deep learning of its artificial intelligence will allow it to learn every day from all the circumstances surrounding driving: the reactions of all the cars around us, the behavior of all the pedestrians we come across, the conditions of all the streets and roads we drive on, traffic movements… With this huge amount of data from all the autonomous cars in the world, situation models will be developed and stored in neural networks, from which new algorithms will be constantly developed and implemented in each autonomous car. In this way, in any driving situation, the on-board computer’s AI will benefit from the experience of millions of situations already experienced, to make the right decision in a fraction of a second.

A thousand and one future applications of AI in automobiles

Today’s diagnostic systems already tell us when to have an overhaul not only by date or mileage, but also by driving habits. This technology will continue to be refined with AI, which will make it possible to make predictions about future problems in our car.

According to a McKinsey report, insurance companies will also be able to use AI to perform risk profiling and premium calculations based on data shared by drivers. Simple AI is already being applied in China to allow drivers to make their own claims reports through an app that guides the user through the entire process.

Artificial intelligence will help predict and avoid traffic jams in the near future by predicting traffic evolution, controlling traffic lights and car flows, for example, from residential areas to large workplaces. It will also be able to control the guidance of navigators in human-driven cars, and of course, that of autonomous cars.

AI will also be used to efficiently manage car-sharing fleets, networks and charging points for electric cars, parking spaces and connections in multimodal mobility systems, among many other aspects of new forms of mobility.

Categorized in: