Sompo Japan Nipponkoa Insurance, Daiichi Kotsu Sangyo and Accenture to Develop Deep Learning Algorithm to Understand Driving Behavior, Improve Safety in Japan

Companies will research how to use data analytics to analyze behavior and potentially reduce automobile accidents

TOKYO; Aug. 25, 2017 – Sompo Japan Nipponkoa Insurance, Daiichi Kotsu Sangyo and Accenture (NYSE: ACN) are collaborating to build a deep learning algorithm using the Intel® IoT Platform Reference Architecture to better understand individual driving habits and identify new ways to transform driver safety within Japan’s transportation industry.

The deep learning algorithm could enable transportation companies to provide personalized safety instructions for drivers, helping reduce the number of accidents, inform the development of optimal driver rosters, and enhance training programs.

Data Flow Overview

Sompo Japan Nipponkoa Insurance will collect data from connected devices installed in Daiichi Kotsu Sangyo’s taxis. In addition to cameras capturing images and telemetry tools recording journey data, biometric information such as heart rates will be collected from consenting taxi drivers through wearable devices.

As part of an ongoing strategic relationship between Accenture and Intel, the data will be processed securely and anonymously using the Intel® IoT Platform Reference Architecture that includes Intel® processor-based servers equipped with Intel®’s high-performance Xeon processors, Intel® Gateway for data collection, and edge computing image processing technology. The data will then be uploaded to the cloud for secure storage and analytics processing.

Accenture will use the input to develop an algorithm that will automatically assess the accident risk for each driver by collating and analyzing images, biometrics, and vehicle data indicating speed and driving behavior. Deep learning, which is one of the emerging advanced analytics techniques available today, will be integral to the data platform.

In an initial Proof of Concept experiment conducted in March 2017 that used data collected from 100 taxis and 100 drivers, the deep-learning algorithm created intelligence that identified signs of drivers’ drowsiness and near-miss accidents from their heart-rate changes and driving behavior.

“Rapid advances in IoT and autonomous driving technologies are bringing new challenges that can only be addressed by using new technologies such as this deep learning algorithm,” said Takuya Kudo, Data Science Center of Excellence global lead and Japan lead for Accenture Analytics, part of Accenture Digital.

As part of this innovative collaboration, Accenture will continue to create new intelligence by applying the latest analytics technologies to address industry challenges. For example, the ability to analyze images on a large, commercial scale is still being developed, and as part of this project Accenture is applying the latest innovations in advanced analytics and data science tools to enable this.