Main menu


8 AI automotive trends in auto industry

8 AI automotive trends in auto industry

AI  automotive utilizations in the auto industry

While the use of AI is being promoted in various industries, the field of use that is closely related to consumers' lives and has the highest expectations is automobiles. Automobiles equipped with AI automotive systems have already appeared on the market, and each country and company is competing with the development toward the realization of high-level autonomous driving. However, looking at the automobile industry, the scope of AI automotive  utilization is not limited to autonomous driving. This time, we would like to introduce the current state of AI automotive utilization in the automobile industry, which is used in various ways.

Typical AI utilization in the automobile industry

First, I will introduce some typical examples of AI automotive utilizations in the automobile industry.

AI automotive: Utilization in automatic driving

The first thing that comes to mind when using AI automotive is this autonomous driving technology. Autonomous driving, in which the system drives the car instead of the driver, is being developed by each country and company with the aim of complete automation.

Autonomous driving is divided into 6 levels from 0 to 5 depending on whether the driving weight is applied to the driver or the system.

At this stage, Level 5 autonomous driving is not in the realization stage in terms of technology and legal development, and research and development of Levels 3 and 4 are underway.Recently, Honda announced in March 2021. It became a hot topic to release the world's first Level 3 commercial vehicle "Legend". We can see that we are definitely on the path of evolution toward the realization of fully autonomous driving.

AI automotive: Utilization in automatic driving

AI automotive: Utilization in the design / inspection phase

When the theme is "automobile x AI", only autonomous driving tends to be in the limelight, but AI utilization in the automobile industry is not the only one. AI automotive  technology is beginning to play an active role in design and inspection.

 Utilization of AI automotive in bonnet design

AI automotive  is also used in the design stage of automobiles. For example, Honda R & D is using AI automotive to design the bonnet design to ensure the safety of pedestrians.

Currently, fatal accidents caused by traffic accidents often damage pedestrians, and are caused by banging the head on the hood at a considerable rate. While it is possible to soften the bonnet to minimize the impact on pedestrians, the bonnet also protects the engine. In addition, improving the bonnet means redesigning the entire design, style, and structure of the automobile, which was designed based on precise optimization calculations, which is not straightforward.

Therefore, the company is working on the utilization of AI automotive  and deep learning technology that aims to predict the optimum shape after considering the overall structure.

 Utilization of AI automotive  for the inspection phase

Image recognition is one of the genres of AI automotive that is good at machine learning and deep learning. Due to the evolution of image recognition technology, AI-based inspection systems are being developed and introduced one after another in the manufacturing industry.

Even in the manufacture of automobiles, by learning images of non-defective products and images of defective parts, AI automotive can distinguish between non-defective products and defective products, shortening the time required for quality inspection, and inspecting with higher accuracy than visual confirmation by humans. Examples of utilization such as performing work have been born.

Utilization of AI automotive  for the inspection phase

AI automotive: Utilization of AI automotive  for car sales

AI automotive  is used not only in autonomous driving, design and inspection, but also in automobile dealers.

Dentsu and Dentsu Digital have developed a unique system that combines the Japanese AI natural dialogue service "Kiku-Hana" with Navitime Japan's car navigation app. We are proceeding with an initiative to have AI automotive  in the in-vehicle smartphone carry out conversations during the test drive, such as test drive route guidance and introduction of car selling points, which the sales staff had been riding with.

This system not only eliminates the sales staff, but also lowers the psychological hurdles for the test drive, and is expected to have the effect of allowing you to enjoy the test drive comfortably. In addition, by digitizing the user's answers to the questions from AI automotive  regarding the test drive, we are aiming to build a system in which the sales staff can utilize the contents in subsequent business negotiations.

AI automotive: Utilization of AI automotive  for car sales

Basic knowledge of AI

Now, let's return to AI and look back on what kind of technology is called AI.

What is AI (artificial intelligence)?

In the first place, AI is a general term for technologies that aim to imitate human intelligence to machines, as you can see from the translation of artificial intelligence. However, AI that replaces humans like the world of science fiction is close to fantasy, and in reality, AI is used as a solution that is effective for specific partial problems. AI developed for a specific purpose in this way is called "weak AI" or "specialized AI", while AI that has not been realized but is versatile enough to replace all human beings is "weak AI" or "specialized AI". It is called "strong AI" or "general purpose AI".

Machine learning (machine learning)

One of the most rapidly evolving technological areas of AI technology is the field called machine learning, in which the computer itself learns and performs recognition, identification, reasoning, and so on.

As the name of machine learning suggests, machine learning makes a computer learn a huge amount of data and analyze and grasp the tendency of the data, even if it is a task that could not be executed unless humans specify patterns in detail. This is a technology that aims to make the machine itself automate tasks.

Image recognition is one of the areas where machine learning is particularly effective. For example, various use cases have been created, such as tasks such as automatically distinguishing and sorting red apples and green apples, and practical tasks such as learning and sorting defective parts. In autonomous driving technology, image recognition technology is also used to detect pedestrians and obstacles through a camera mounted on the vehicle body and to recognize the driving environment.

Machine learning (machine learning)

Deep learning

Deep learning is a technique included in machine learning. It is a system inspired by the structure of neurons in the human brain, and is characterized in that it can be learned without the need for humans to specify the "features" that are normally required in machine learning.

At automobile production sites, it is expected to be used for discovering trends that even skilled engineers and designers cannot find, such as discovering the relationship between behaviors common to drivers and equipment failures.

 AI automotive , 8 examples

Finally, we will introduce actual AI automotive  utilization cases, including our case.

AI automotive recommendation of destination through analysis of latent needs

When you go out to play somewhere by car, you need to search for your own destination. However, if the destination is not clear, such as "I want to drive somehow", it is difficult to verbalize the needs, and as a result, there is a limit to searching.

At Laboro.AI automotive , in research and development with a major automobile manufacturer, we worked on the development of an AI automotive  recommendation system that analyzes the potential needs of users through dialogue with AI automotive  and proposes destinations based on the contents. By utilizing big data such as word-of-mouth data on the Web related to tourist spots and map information as data, and making recommendations based on dialogue feedback, users themselves can propose destinations that meet latent needs that are difficult to verbalize. We are developing it as a new mechanism to do.

AI automotive: Toyota's autonomous driving development

In 2018, Toyota Motor Corporation established a new subsidiary, Toyota Research Institute Advanced Development, which aims to realize self-driving cars. We are developing a system in which AI automotive  performs the generation of highly accurate maps and the judgment of the situation during driving, and we are proceeding with the development in cooperation with Amazon and others.

AI automotive: Tesla's self-driving car development

Naturally, the AI automotive  ​​system installed in automobiles is powered by electricity, so it is said to be compatible with electric vehicles. Tesla, the world leader in electric vehicles, is also enthusiastic about developing self-driving cars and has already put high-level driver assistance systems on the market.

AI automotive: License plate reading

Here is an example of using AI automotive technology to read the license plate of a car. After taking a picture of the license plate part of the car and performing image processing, it is read by AI automotive  who learned about character recognition, and the information is automatically accumulated, and in addition to vehicle management, it is a violation. It is expected to be used for cracking down on vehicles.

AI automotive: Distance estimation using an in-vehicle camera

In order to realize autonomous driving technology and driving support systems, it is necessary to measure the distance to the surroundings. In this case, a system has been developed that can estimate the distance by combining two cameras two-dimensionally, and can also estimate the distance with only one camera.

(Source: Car Watch "Albert, announced at GTC Japan 2017, a depth estimation (distance estimation) engine that utilizes deep learning that can also be used with monocular cameras" )

AI automotive: Improving inspection efficiency in the press process

Audi, an automobile manufacturer, is making efforts to establish a mass production system in the factory by utilizing machine learning technology to automatically recognize cracks and scratches on metal plates that occur during press working. In recent years, automobile design has become more sophisticated, and the quality standards that are inevitably required are also increasing. At Audi, we are able to inspect all the parts processed at the press factory on the spot, which has greatly improved operational efficiency.

AI automotive: Taxi demand forecast

Surprisingly, AI automotive  is also used to forecast taxi demand from the perspective of the same car. If you are a veteran driver, you can understand the area and time zone where passengers are likely to gather by experience and intuition, but it is often difficult for new drivers. Considering the time of day and the weather, there are examples of utilization such as predicting areas where passengers are likely to gather and recommending to drivers.

AI automotive: Determining the optimal route for a shared bus

Osaka Metro and Osaka City Bus have started a social experiment to optimize the route of shared buses using AI automotive . It tells the time and destination that the user wants to use, makes a reservation, aggregates users with similar wishes, and uses AI  automotive to calculate the optimal operation route. Unlike a fixed-route bus, which has a fixed time and operating route, it can be said to be an example of utilization unique to a shared bus that requires the time to be set according to demand and the shortest distance to the destination. In the experiment, a total of 10 small bus (8-seater) drivers are supposed to patrol according to AI automotive  instructions, and are expected to be used for traveling to and from stations, hospitals, supermarkets, etc.