What are the Levels of Autonomous Driving?

In order to understand the differences in autonomous driving, let’s take a look at the “Levels of Driving Automation” as defined by the American Society of Automotive Engineers (hereafter referred to as SAE), which is currently accepted as the international standard. SAE classifies driving automation features into five levels as follows.

At Levels 1 and 2, a human driver is the main operator of the vehicle, whereas at Levels 3 to 5, the vehicle itself, that is, the system, becomes the main operator.

From among these levels, SoftBank aims to implement autonomous driving vehicles with Level 4 or higher in Japan.

At Level 3, if it becomes difficult for the system to continue operating, the driver has to swiftly take over the driving operation. On the other hand, at Level 4, although it is only under certain conditions such as in specific areas, advanced autonomous driving is required, such as safely stopping the vehicle without the driver driving, even if a problem occurs and the system is unable to operate.

Therefore, the achievement of autonomous driving at Level 4 or higher can contribute to solving social issues in Japan, such as driver shortages and an increase in elderly people with mobility difficulties due to the return of their driver’s licenses.

Global Landscape Surrounding Autonomous Driving

Autonomous driving technologies are being actively developed across the world. America is one of the representative countries in this field, with companies, such as Waymo, Cruise, and Aurora Innovation being particularly well known for their advanced technological development. These companies have already made practical use of autonomous driving technologies that are equivalent to Level 4, such as offering driverless autonomous taxi services on public roads.

Advanced technological development is also gaining momentum in other countries such as China, where similar services are operated.

Autonomous Driving in Japan

The majority of autonomous driving systems currently being used in Japan are equivalent to Level 2 technologies, which can only be used under certain conditions.

To advance the implementation of autonomous driving technology in Japan, the following five challenges will need to be solved.
(1) Unreadiness of autonomous driving technology
(2) Lack of social acceptance (autonomous driving is still not accepted by society)
(3) Regulation
(4) High operation costs
(5) Lack of established business models

However, with regard to “(3) Regulation”, there has been movements toward the adoption of autonomous driving in Japan, such as the inclusion of provisions related to the legalization of Level 4 autonomous driving in the revised Road Traffic Act, which is expected to come into effect in fiscal year 2023.

The Research Institute of Advanced Technology’s Initiatives

As part of the research and development towards the implementation of autonomous driving technology into society, the Research Institute of Advanced Technology has built an autonomous driving operation platform and are pursuing the following three initiatives.

1. Building an autonomous driving operation system (to resolve issue “(4) high operation costs”)

Extensive operation costs are required today in order to realize autonomous driving. Efforts are also required to address various tasks that drivers have been responsible for until now. Therefore, it is not easy to operate a sustainable autonomous driving business.

This is why the Research Institute of Advanced Technology is conducting activities to automate the tasks towards the practical application of autonomous driving and improve efficiency, by aggregating the data from the vehicles, in-vehicle conditions, and transportation infrastructure, on to our autonomous driving operation platform and analyze using AI.

We are currently working to create a system that enables a single operator to operate multiple vehicles during the remote monitoring that is necessary for the operation of Level 4 autonomous driving vehicles.

2. Creating a system for improved safety (to resolve issue “(1) unreadiness of autonomous driving technology” and “(2) lack of social acceptance”)

When operating an autonomous driving vehicle, it is necessary to set an “Operational Design Domain (ODD)” that is appropriate for the autonomous driving system.

For this process, data such as the traffic flow of people and vehicles in the planned driving area is collected and digital twin^※ is used. We conduct R&D by verifying various driving situations in advance using digital twins, creating an ODD, and then actually driving the vehicle.

• ※Digital twin: A technology that replicates data collected from real world objects and environments in a virtual space.

Furthermore, data is collected for each example of problems that arise during the research and verification test for autonomous driving technologies. Scenarios are then created so that this can also be utilized by other companies working on the development of autonomous driving systems. The Research Institute of Advanced Technology strives to contribute to the improvement of autonomous driving technologies across Japan as a whole in addition to our own research.

3. Working on the MaaS initiative to solve business model problems (to resolve issue “(5) lack of established business models”)

Transportation businesses that use autonomous driving vehicles, which are expensive in comparison to ordinary passenger cars, require innovative business models.

“MONET Technologies”, formed by SoftBank and Toyota Motor Corporation as a joint investment, is working towards the creation of new mobility services and MaaS businesses.

There is still no precedent for implementation of autonomous driving Level 4 or above in Japan. However, through the accumulation of various initiatives, its realization is gradually approaching.

SoftBank will continue its R&D activities towards the realization of an autonomous driving society with the aim of solving social issues related to mobility and transportation.