Telematics

Telematics is a field dedicated to collecting, transmitting and analysing data about vehicles and their movement.

Our telematics systems gather and analyse behavioral, contextual and driving data to provide innovative solutions for our business partners.

Our Approach

Data collection
Data enrichment
Hazardous events detection
Behavior analysis
Data visualization
Feedback
Gather telematic data from various sensors. This can include GPS, accelerometer from a mobile phone or tag, OBD2 dongles or even cameras and radars.
Add contextual information, such as road types, speed limits, real-time traffic and weather conditions from external providers.
Identify driver’s actions, such as speeding, harsh accelerations, riding on flat tires, peaks in engine RPMs or following another car too closely that are a sign of bad driving habits.
Build driver’s profile and evaluate each driver in terms of safe and economical driving.
Present collected data and results of the analysis in form of clear dashboards and reports.
Provide the driver with feedback and tips what she can improve.

Different needs, different approaches

We offer three different levels of telematics solutions, each of them suited best for different use-case:

Entry level

Data is collected from sensors in the driver’s mobile phone (such as GPS and accelerometer). The solution is easy to introduce, even on a mass market (as no additional hardware is required), but it offers only limited precision of the analysis. No car diagnostic is available. The precision of acceleration detection is lower than in case of high precision telematics.

High precision

Data is collected both from sensors in a phone and from additional devices that read diagnostic information from the vehicle. This approach allows us to include much more parameters in the analysis, resulting in more accurate drivers’ profiles, but it requires additional hardware to be installed in each vehicle.

Advanced

In our next-generation telematics solutions employ sophisticated state-of-the-art machine learning algorithms and additional sensors (such as a camera or radar) to achieve superior analysis precision.

Solution Architecture

Sparkbit builds end-to-end telematics platforms. We deliver the following key software components:

Mobile application

for the driver that gathers the data, transmits it to the server for processing and visualizes the results of the analysis. When additional hardware sensors are used, the mobile application can also act as a bridge between them and the server.

Server

highly available, horizontally scalable system that securely stores, processes and analyzes the telematic data. The server component can be offered either as a cloud-based solution or can be installed in client’s infrastructure.

Web portal

that provides insights into performance of each asset within the fleet as well as profiles of each driver.

Business Value

Telematic systems created by Sparkbit provide additional business value for various industries.

Insurance

Use telematic data and driver’s profile to offer usage-based insurance (UBI), where the policy price is calculated individually for each driver based on his/her driving style and skills. UBI promotes good driving habits and in the same time helps the insurer to reduce the financial risk and cost of handling claims.

Logistics and fleet management

Telematic data helps to improve fleet utilisation and can be also used to lower fleet maintenance costs, when combined with adaptive predictive maintenance systems.

Telematic systems provide also great benefits to the society:
Improved road safety

Telematic systems detect hazardous behavior on the road. They give drivers feedback on their driving style. It has been observed that telematic-based insurance products encourage people to drive safer and more carefully, reducing the number and severity of road accidents.

Economical driving

Driving style evaluation algorithms promote environmentally friendly driving and thus help to reduce air pollution and noise.

Sparkbit Telematics

Telematic platforms created by Sparkbit are used by drivers around the world. We delivered solutions and algorithms for leading telematics providers.

330 million kilometers of trips (more than to the Sun and back) made by users of our applications
2 million trips each month
Drivers all around the world
30 000 active users each month
Deployments for clients from Italy, Canada, Poland and France

Case Studies

OctoU

System for Analyzing Driving Behavior of Users
Challenge

Our client- Octo Telematics - sought to build a powerful platform to evaluate driving behavior of users based on data collected by mobile applications. They had only a high-level vision of the product with no specific functional requirements and no clear expectations as to number of users or amount of data each user will generate.

Our challenge was to design and implement horizontally scalable, reliable and fault-tolerant system as well as prepare algorithms for detection of events in raw data and scoring users based on data collected from various sources.

Solution

Without clear requirements we were forced to first quickly create an MVP and receive customer feedback. After that we were adding new features in iterations between 2 and 4 weeks. The system went live after 1 year of development.

We developed a feature-rich system that allows users to measure their driving skills. The system is based on heterogeneous data, such as real-time GPS data, weather and road conditions, traffic congestion, speed limits and drivers’ lifestyle information.

Each users trip is assigned a score that describes how safely the user was driving. The algorithm detects hazardous events, such as speed limit violations, harsh acceleration, braking, and cornering.

The system also contains social features, such as commenting and liking friends’ trips, collecting badges, receiving tips etc. The system is distributed and is deployed on multiple nodes that asynchronously process incoming data.

Tools
  • Core system:
  • Java 8
  • Spring
  • ActiveMQ
  • MySQL
  • Cassandra
  • Web console:
  • Angular 5
  • TypeScript
June 2014 - still being developed.

IFC

Octo U Customization for a Canadian Insurance Company
Challenge

Our client’s client - the largest provider of property and casualty insurance in Canada - wanted to introduce Usage-based insurance solution relying on smartphone apps to their existing software ecosystem.

Our challenge was to handle much heavier load in the system, provide production support and quickly react to issues and failures.

Solution

As the core platform is built on top of a horizontally scalable NoSQL database, we were able to easily adjust the systems capacity in accordance with the increasing traffic. We have also designed the required infrastructure setup necessary to support the expected volume of data.

The system is deployed on over 20 nodes with total of 300 CPU, so we had to introduce a centralized log analysis mechanism. We have implemented it using Elasticsearch and Kibana. We have utilized this mechanism to identify and eliminate several performance bottlenecks in the system.

Tools
  • Core:
  • Java 8
  • Spring
  • ActiveMQ
  • MySQL
  • Cassandra
  • Monitoring:
  • Elasticsearch
  • Kibana
June 2016 - December 2016, currently in maintenance.

go11

System for Insurance companies to provide UBI to customers
Challenge

One of the leading Polish insuretech company focused on car insurance approached us with an idea to create a platform that would allow insurance companies to gather data about their customers’ car trips to provide a Usage Based Insurance model using the users smartphones.

This called for creating a performant system that would be able to analyse driving behaviour of thousands users each day integrating with external systems. Additionally our partners goal was to have an end-to-end solution accessible from a wide range of devices.

Solution

We have implemented the platform as a distributed backend system with a robust and scalable API that is used by mobile applications that we also developed.

The driving behaviour of users is analyzed and scored using data from sensors such as GPS, accelerometer, gyroscope and digital compass enriched by contextual data like current weather conditions, road types and traffic congestion that we access using external providers. The end-user can browse through their driving history and see each trip on a map with hazardous events highlighted to provide feedback. Apart from the mobile applications two web applications were developed for both end-users and insurance operators.

Tools
    Backend: Java 8, Spring Boot, MySQL, Cassandra, ActiveMQ
    Web Application: Angular2
    Android: Java, Android SDK, Gradle
    iOS: Swift 4, CocoaPods, PromiseKit
    Infrastructure and Monitoring: Influxdata TICK, Crashlytics, AWS EC2, Jenkins, Ansible, fastlane
June 2017 - still being developed, production launched in December 2017. Currently in beta tests.

HUMN

Telematic system for fleets
Challenge

Our client wanted to include in his offering a telematics platform that would be offered to taxi fleets. The goal of the system was to evaluate driving behavior of users based on data collected from vehicle’s OBD2 interfaces. For other purposes, the client wants to collect also high-definition pictures from a dash camera. The score history should be also stored on the blockchain, to be used as an input for smart contracts that secure discounts in insurance rates for good drivers.

Our challenge was to design a fully scalable platform capable of handling the stream of both telematic and image data.

Solution

We have implemented a system based on big data stack: Kafka, Scala, Spark Streaming and Cassandra. The score history is additionally persisted in BigChain DB and the transaction references are further passed to Ethereum Swarm.

As the client didn’t want to be tied to one particular cloud provider, our solution is fully dockerized and deployed and managed through Kubernetes.

Tools
    Core system: Scala, Kafka, Spark Streaming, Cassandra, Docker, Kubernetes
    Blockchain component: BigChain DB, Swarm, Tendermint
    Web console: React, TypeScript
April 2018 - still being developed.

Telematics in Japan

Challenge

Our client, a leading consultancy firm in Japan, wanted to extend their offer by building a telematics platform that evaluates users’ driving behavior. The system had to be modular and customizable, so that it can be offered in different configurations to clients in various sectors.

Key non-functional requirements included full horizontal scalability and low operational costs.

Solution

Sparkbit created an end-to-end solution, consisting of mobile applications that gather telematic data, distributed backend platform and a modern web dashboard. To achieve the required scalability and fault tolerance, we have used Kafka as the main messaging system, Cassandra as the primary data store and Redis as a distributed cache. The platform is built in Kotlin, using Spring 5 framework. The platform is deployed on AWS and utilizes multiple services such as RDS and ECS.

We have based the cartographic data on open-source data, reducing this way costs associated to using commercial APIs.

The mobile applications are modular and consist of a telematic SDK and a UI layer. This way the SDK can be offered as a standalone component that can be embedded in other applications.

Tools
    Backend platform: Kotlin, Kafka, Cassandra, Spring 5, Redis, Docker, AWS EC2, ECS, RDS, S3
    iOS: Swift
    Android: Kotlin
    Web console: TypeScript, React
June 2018 - still being developed.