Changing Driver Behavior through Technology and Insurance Systems

"GOOD DRIVE" – An Insurance Service to Reduce Accident Risk

Feb 24, 2021

Is it possible to reduce accidents through car insurance? GOOD DRIVE, an automobile insurance system provided in Japan that is linked to driving behavior, wants to find out. As of May 2020, this system has also been linked to the smartphone application “Yahoo! Car Navigation” provided by Yahoo Japan Corporation, and we are making progress toward the realization of a safe society without traffic accidents.

  • Eisuke Ishii

    Direct Marketing Dept.
    Marketing Div.
    Sony Assurance Inc.

  • Chihiro Umemura

    Contents-Marketing Planning Dept.
    Marketing Div.
    Sony Assurance Inc.

  • Koji Arakuta

    Direct Marketing Dept.
    Marketing Div.
    Sony Assurance Inc.

Cashback on automobile insurance with accident risk calculated by advanced technology

──What kind of automobile insurance is GOOD DRIVE?

Eisuke Ishii:GOOD DRIVE’s smartphone application is used to measure and score driving behavior, with cash back insurance premiums of up to 30% awarded to drivers with low accident risk. The score is calculated based on factors such as acceleration, braking, steering, and smartphone use while driving. The driver can check their current score on the app, see what the score is based on, and get advice on how to improve it. If fully autonomous driving becomes more widespread, the risk of accidents should decrease considerably. However, in countries and regions such as Japan with networks of narrow roads, autonomous driving is much more challenging, and it will take considerably more time for the accident reduction effect to become noticeable. Until then, I think that this product, which can reduce the number of accidents by changing driver awareness and behavior, is of great significance.

──How exactly does the measurement and analytics work?

Ishii:To use the service, you first attach a special Bluetooth® device to your vehicle’s accessory socket and install a special app on your smartphone. It works in conjunction with the engine to activate the app, so when you turn on the car, it will automatically start measuring driving data. The measurement is done by the app. Measurement data obtained from the accelerometer, gyroscope, and GPS in the smartphone are collected in the cloud computing system, and accident risk is calculated with a prediction model created in association with accident data held by Sony Assurance. AI algorithms developed by Sony’s R&D Center are used in both the app and the cloud to measure driving characteristics and estimate accident risk.

The device is attached to the vehicle's accessory socket, and measurements are taken using a dedicated smartphone application.

Ishii:The device is also equipped with an emergency button, and in the event of an accident, the app displays emergency contact information and allows you to place a call immediately. Location and contact information will also be linked to Sony Assurance’s accident receptionist for a smoother response.

Specifications tailored for driver convenience

──Automobile insurance and technology to measure driving data already exists, so what makes GOOD DRIVE special?

Ishii:As an insurance product, its main feature is that it focuses on preventing accidents and changing driver behavior as opposed to the usual type of insurance that helps you after an accident. By encouraging safe driving, customers are less likely to be involved in accidents and also benefit greatly from lower insurance rates. For us, it is also more profitable because we pay less in premiums. In addition, we contribute to society as a whole by reducing the number of accidents. In other words, it’s an ideal system where everybody wins, and we put a lot of effort into realizing it.

Koji Arakuta:Technically speaking, the aforementioned automatic start-up and measurement, and the fact that the analysis is done with a smartphone, are the major differences from other driving data measurement services. When we first developed the app, drivers were required to launch it every time they got in the car, but that would have been a hassle, and they likely wouldn’t have used it on a daily basis. No matter how good a service, if it’s not user-friendly, it won’t catch on. So, after repeated discussions, we developed the app in its current form.

──If I use my smartphone for measurement, will I have to fix it somewhere in the car? Also, is there any chance that the score will be lowered if I move it?

Chihiro Umemura:While driving, the app measures driving data regardless of where the smartphone is placed or what model it is. It doesn’t matter if you leave it in your seat, drink holder, bag or pocket. And, for example, if you try to pick something up while the car is stopped and your smartphone is in your pocket, it won’t recognize it as a risky action. That is because our in-house developed algorithm incorporates an enormous amount of data collected from actual driving tests to find out which behavior is risky.

Driver characteristics are visualized on the app.

Communicating in a way that leads to changes in behavior

──How did you go about the development process?

Ishii:GOOD DRIVE was developed using assets from three different Sony companies: Sony’s AI technologies, cloud computing, signal processing, and sensor technologies; Sony Assurance’s insurance service know-how and data; and Sony Network Communications’ software development knowledge. As a result, we have developed a product that accurately and precisely measures risky behavior and is also very convenient for drivers.

Umemura:The project was carried out by a diverse group of people from different industries and backgrounds, including people from the financial industry and the engineers who were actually involved in the development. However, since we shared the same desire to “create an excellent service for our customers,” we were able to have lively discussions on how to create the system.

Arakuta:There were planning members, sales members, and engineering members, and usually each group has its own set of requirements, but everyone involved in this project was very passionate, so we were able to share a lot of opinions about what we could do to benefit the customer.

──I’m sure that resulted in many new ideas and functions.

Umemura:In addition to the automatic start-up and system that doesn’t rely on the smartphone being in a fixed position, we discussed how to communicate the score obtained from the analysis to the driver. If we simply told drivers the results, it would be boring. And some people would give up, saying, “I just can’t drive well.” However, those are exactly the kind of people we want to help. Therefore, we decided to specifically tell them which driving maneuvers they aren’t good at and how they can improve next time.

Arakuta:We also thought a lot about the timing. If you suddenly hit the brakes while driving and the app says, “That was dangerous,” you would probably get irritated at the obviousness of it. Even worse, the message could be distracting, causing the driver to look at their phone or drive even more recklessly. As such, we settled on saving the advice until after the driving is done.

Umemura:When designing the app, we focused on making it attractive. Otherwise, people wouldn’t look at it and thus wouldn’t have a chance to learn about their driving. We want users to keep checking it and keep trying to improve.

The driving score display screen. In addition to advice on how to improve one’s driving score, driver’s score is represented by a slowly rotating circle. The goal is to make it a natural part of our customer’s lives and to get them to unconsciously improve their driving.

Accident risks reduced by 15%
Insurance that makes driving more fun

──I heard that you conducted an experiment with Sony Group employees. Could you tell us about it?

Arakuta:The experiment was conducted over a period of two months with a total of 600 participants. First, we divided the participants into two groups, Group A and Group B. For Group A, we simply asked for recording their driving time. For Group B, we likewise recorded their driving time data in the first month, but from the second month, we told them that we would calculate their driving scores based on our algorithm and show them. We also promised them a reward if they improved their scores. Two months later, we determined that accident risk was reduced by about 15% in Group B compared to Group A, thus demonstrating GOOD DRIVE’s ability to reduce accident risk.

──Given the results of this experience, what are your hopes and expectations for the future?

Arakuta:Naturally, we were very happy with the results of the experiment, but I was also delighted that some of the participants reported that everyday driving became more fun as a result. I was worried that having their driving rated would make them nervous, but the response was very positive. Still, we’re not going to stop here. We will continue making improvements so that things get even better. And by spreading GOOD DRIVE, we hope to contribute to not only the safety and security of the driver but also to society as a whole surrounding mobility.

Related article