The Future

We were inspired to begin the development of the Subaru Global Platform when we noticed that car measurement data did not match the feelings of the passengers in the cars. This led us to reconsider our measurement precision and measurement points, to develop a measurement method whereby the data and feelings would match. This materialized into the Subaru Global Platform. We further advance our research considering human body structure. Some of the results have already been incorporated into recent Subaru cars, but as we continue our research, we will reflect more results in our future cars. By evolving the Subaru Global Platform further, we hope to improve the peace of mind that our drivers feel and the enjoyment of controlling Subaru cars. It will also be a key in the autonomous-driving era. We accelerate to enhance this as an important core technology.

Why does Subaru have a low fatality rate? We believe that this is a result from the vehicle developments based on our philosophy of All-Around Safety. There are four aspects in Subaru’s unique All-Around Safety philosophy: primary safety, driving safety, preventive safety, and passive safety. “Primary safety” means factors such as excellent visibility and fatigue-avoiding packaging. “Driving safety” refers to accurate avoidance performance of objects on the roads and controllability after the avoidance. “Preventive safety” is performances to prevent accidents and to mitigate damages utilizing the features such as pre-collision braking of EyeSight. If an accident occurs even with these three-step measures, we protect passengers in “Passive safety” phase. Third-party evaluations mainly focus on passive safety. Some are starting to cover preventive safety as well. But in Subaru, we consider all four safety aspects, and we think that is why we have lower fatal accident rate in the real world. Henceforth, we plan to continue to enhance our work in each of these four important areas, and to add connected safety. We aim to achieve our goal of zero fatal accidents by 2030 by utilizing AI technologies.

These charts show analyses results of the factors in fatal accidents in the US and fatal and serious-injury accidents in Japan in 2017 involving Subaru cars. The blue parts show accidents caused by own cars, and the red parts show accidents caused by other vehicles. Both in Japan and in the US, 60 to 70 percent of the accidents were caused by own cars.

To achieve zero fatal traffic accidents by 2030, first of all, we seek a 65-percent reduction in fatal traffic accidents by enhancement of advanced driver-assistance systems (ADAS). For the remaining 35 percent, we plan to reduce extent of injuries of passengers through continual improvement of passive safety, a Subaru specialty. We also plan to save more lives with advanced automatic collision notification (AACN) using connected services. We expect the combination of these measures will lead us toward achieving zero fatal traffic accidents by 2030.

Regarding advanced driver-assistance systems, very effective in reducing accidents, we will evolve Subaru’s unique EyeSight technology to the next generation. We continue to evolve our system, expanding its response to accidents on local roads, for example, and thoroughly hone accident avoidance and driving support technologies.

EyeSight has focused on avoiding crashes in straight line driving, but we are now enhancing its protection areas for intersection and city accidents (such as hook accidents involving pedestrians, head-on crashes with oncoming cars during a right turn, and right-angle crashes). We also enhance the system to monitor the state of the driver and to cover the operational errors. We will introduce features to prevent lane departure by recognizing the edge of the road even without white lines and to stop the car safely if the driver loses consciousness.

In the next generation of EyeSight, we are combining the stereo camera with a high-precision map locator and four radars in front and rear, in order to hone our driver-assistance technologies that are categorized as Level 2 automated driving such as lane change assist, curve-predicting automatic deceleration, and traffic jam hands-off assistance. We hope to launch cutting-edge features that can realize comfortable and safe highway driving with mass-market cars.

EyeSight is structured to measure distances from the parallax between two cameras. If there is, for example, a car 10 meters ahead, EyeSight recognizes it as a cloud of about 4,000 distance data points. This structure allows for more fine-grained control. EyeSight also excels at road shape detection, object position tracking and movement prediction. Thus, the stereo camera system alone is enough to collect the information needed for recognition and judgment efficiently.

We evolve EyeSight performances to the next level by adding AI judgment ability to the recognition capability. We can improve the capability to detect drivable areas with the fusion of Stereo camera and AI. For example, one is driving on a snowy road where the center line and road shoulder are completely obscured, AI can calculate drivable areas and figure out the driving route. Also, if a vehicle in front brakes suddenly and the AI judges that it is impossible to stop, it can find the optimal evasion route and control the car. Through our research and development to utilize the potential of EyeSight in such ways, we will continue to upgrade its recognition and judgment capabilities and enhance safety on all kinds of roads.

Subaru has adopted a Driver Monitoring System (DMS) since the launch of Forester in 2018. The technology not only recognizes individuals, but also detects driver’s inattentive driving and drowsiness and alerts the driver. We will continue to link up a Driver Monitoring System (DMS) with various kinds of control systems including EyeSight. Also, in 2019, Subaru has released a new infotainment system to the North American market. The amount of data generated by cars is increasing, and so are inputs from such as connected services, as well as things to control. Thus, communication between passengers and cars is becoming more important. Our new system is the first in the world to implement a framework in mass-produced cars that drives important devices such as meters and infotainment features like audios and navigation all in one unit, while taking care of security and reliability. We plan to roll it out globally.

Subaru has been providing the connected services, Subaru Starlink, in the US since 2016.

This represents the state of subscription to the Subaru Starlink service in the US as of November 2019. Many purchasers of new cars subscribed: 96 percent in the free period for the first three years, an average of 48 percent for paid service. We have in mind to roll this service out in Canada and then in Japan.

As one element of our connected services, we are rolling out advanced automatic collision notification (AACN) to the Japanese market. This system reports various kinds of data collected automatically to a call center in the event of an accident. It uses an algorithm to estimate the probability of death or serious injury and judges whether an ambulance is needed. If the system judges that it requires an ambulance, it calls for one. Research indicates that fatalities can be reduced by 70 percent if a rescue starts 10 minutes earlier, so we think the system will be important for achieving zero fatal accidents by 2030.

We hope to expand our AACN features even further. We have plans to improve the information precision of our AACN to push up lifesaving rates in accidents: for example, by reporting the number of passengers in the car, by linking with EyeSight to report on pedestrian involvement, or by using the Driver Monitoring System (DMS) to judge and report if the driver is acutely ill.

Another key point in achieving zero fatal traffic accidents is mitigating damages by improving passive safety performances. Subaru introduced Japan’s first pedestrian protection airbag in the Impreza released in 2016, but fatal traffic accidents involving cyclists are also on the rise. Our current pedestrian protection airbags have trouble protecting the heads of the cyclists. Therefore, we are studying airbags assuming a crash with cyclists. Also, in regard to passenger protection, our studies have been utilizing passenger injury indices of adult men. But when we looked at actual fatal accident cases, we saw that many cases were involving elder people with weak bones. When we considered how to save them, we have figured out that the keys may be the timing of inflating the airbag and the control of the restraining force of the seatbelt. Now our plan is to further study how fast the airbag can be deployed with EyeSight linkage and how to protect passengers from breaking bones by controlling retaining force of the seatbelts considering ages of passengers monitored through Driver Monitoring System.
In this way, we pursue continual enhancement in the three areas; ADAS evolution, AACN utilization, and passive safety, to realize zero fatal traffic accidents by 2030.