CASE STUDY

AUTO UNION

One of the top World global automotive suppliers

Solution

Scaling internal IT department

Industry

Automotive electronics

YEAR

2018

PROJECT DESCRIPTION

The AUTO UNION Group is among the Top-40 automotive parts suppliers in the world with over a hundred years of development and manufacturing or lighting and electronic components. With over 7 billion Euro in sales in fiscal year 2017/2018 and 40,000+ staff member spread across 35 countries, AUTO UNION is at the frontier of research and development. The company engineers electronic solutions that increase energy efficiency, minimize CO2 emissions and enhance driver and passenger safety and comfort.

Since the client’s primary expertise lies with hardware development, testing, and manufacturing, Computools experts were invited to join the distributed team working on an Artificial Intelligence component for an autonomous-driving system. Computools software engineers helped establish and train an AI system to increase the fidelity of pedestrian detection.

Inefficient technology stack choices for Autonomous driving system’s database and Front End

Demanding development schedule with limited time for internal IT department scaling

Overcomplicated Front End features that hinder system performance

Market-leading manufacturer with strong global demand for its products

Hardware and software development fusion to ensure stable and safe autonomous driving capabilities

Culture of consistent investment in innovation aimed at consolidating market positions

Structured programming approach that enhances development clarity and quality

Business Process Mapping

Described

The AUTO UNION Group experts performed Business Process Mapping. The existing Autonomous driving system required a major upgrading to ensure a necessary pedestrian detection fidelity level. Computools engineers received a detailed system requirements specification with a list of functional and non-functional features to be implemented in the course of AI-based system development.

Modeled

Upon business process modelling analysis, Computools experts identified the critical weaknesses within the existing system and suggested architecture improvements and an alternative technology stack. The approved development strategy ensured seamless performance of the Autonomous driving system software upgrade.

Planned

To meet the tight timeline, Computools experts designed a fast-paced development plan. The majority of large-scale architecture planning was performed on the client’s side, with our engineers providing technology-specific insight and expertise. The project’s timely implementation was the product of cohesive efforts of the in-house and outstaffing specialists.

Digital Platform & Technology

Computools
Team Lead
Software Engineer
CLIENT
Project Coordinator
Software Engineer
Team Lead
Communication
CouchDB

CouchDB enables fast data indexing and retrieval, while the changes feed provides updates. The use of JSON-based document format accommodates a variety of languages that come with multiple libraries. CouchDB is suited for a large and complex application, as is the case with Autonomous driving system for AUTO UNION, as the database is flexible and scalable.

RabbitMQ

To support video upload and cutting for Autonomous driving system, Computools engineers implemented an open source message broker. RabbitMQ supports a variety of messaging protocols and enables reliable asynchronous processing and flexible routing through exchanges before the messages arrive at queues. The RabbitMQ’s tracing feature is invaluable during development and debugging, as it allows software engineers to locate problems and errors.

Microsoft Azure

Microsoft Azure is the top choice for enterprise-level cloud computing service. Beside speed and agility, this solution ensures data security and superior recovery in case of unpredictable failures. Unlike other cloud services, like AWS, Microsoft Azure enables a hybrid environment that balances control and flexibility of a combined on-site and cloud infrastructure.

FFMpeg

Autonomous driving system relies, among other sensors, on video input that needs to be processed quickly and reliably to ensure a high fidelity of pedestrian and obstacle detection. Computools software engineers suggested the implementation of the open source FFMpeg encoder. This solution secured high-performance handling of all video and audio file formats without the need for codecs.

Specification

Computools specialists assisted the AUTO UNION Group in-house IT department in identifying the functional and non-functional requirements for an upgraded Autonomous driving system. Our engineers’ technological expertise coupled with the client’s experience in automotive electronics software development enabled the distributed team to map out the project’s implementation timeline.

Goal 1. Check that Serial Number is not already registered and is eligible to participate in the program. 2. Get printer model. Entry points 1. Register Printer (the user is not logged in). 2. Add Printer (the user is logged in). Post conditions 1. Serial Number check is passed or failed. Main Flow 1. The user provides S/N.

COMPUTOOLS TEAM

To satisfy the client’s demand for technological expertise with AI-powered systems, as well as CouchDB, Docker, and MicrosoftAzure, Computools screened the talent pool and suggested a number of experienced software engineers and QA specialists to join the distributed team.

BACKSTAGE

To keep the development on schedule, Computools software engineers rely on an established communication framework. Routine meetings, interspersed with brainstorming and feedback sessions keep all members of the distributed team in the loop, enhance understanding, and accelerate development.

Dev Meeting
A staple of the chosen development methodology, daily scrums ensured all members of the distributed team performed their assigned tasks according to the sprint plan. Occurring impediments were promptly addressed by the Team Lead and the Scrum Master.
Planning Session
Starting sprints with planning sessions enabled the team to assess the workload, decompose large problems, and assign software engineers with their respective tasks. This approach allows the team to keep up the rapid development pace while addressing arising challenges.
Feedback Session
All team members participated in the end-of-sprint feedback sessions and discussions. Exchanging ideas and opinions between Computools and client-side engineers inspired innovative solutions and increased development efficiency.

Internal Communication Flow

To ensure transparent communication and understanding among the members of the distributed team, Computools applied an established communication framework. Tasks were assigned with Jira, Email was used to deliver status reports and updates, while daily meetings were held via Skype video and audio.

Project Management Methodology

The Scrum approach was chosen for this project to secure fast-paced development and cohesive collaboration across the international distributed team. Daily scrums were held to assess the development progress and address arising challenges. Scrum Master assigned tasks, tracked their progress and managed resources through Jira while Computools’ Project Manager enabled seamless communication and strict schedule adherence by our software engineers.

PROJECT TIMELINE

1 week
2 weeks
6 months
1
Background
First Sight
  • Client's business processes analysis
  • Identifying Strengths & Weaknesses
  • Setting development goals with the client
  • Deriving specification, tasks & deadlines
2
Strategy
Planning
  • Business Process modelling
  • Technology selection
Project Management Flow
  • Expertise-based talent evaluation and team formation
  • Internal communication flow establishment
  • Project management methodology
Development Flow
  • Development Methodology
3
Implementation
System Implementation
  • System requirements’ analysis
  • Roles and tasks assignment
Software Development
  • Code Creating & Refactoring
  • Testing & Fixing
Launch
  • Verification of safety requirements
  • System Launch
  • Deploy support

CODE DEVELOPMENT

Sprint-planning sessions ensured continuous delivery of the project. For each task assigned through Jira, Computools engineers created an individual feature branch. Upon successful code review by the Team Lead, the branches were merged into Dev. QA engineers performed automatic and manual testing before the build process. This approach ensured the project was stable and error-free and complied with the client’s requirement specification.

CODE SAMPLE

Scan QR code to find code sample

Or use link below

https://github.com/FedYura2/samples

Outcome

As a part of an international distributed team, Computools engineers used their expertise to upgrade the AI-powered Autonomous driving system and ensure the precision of obstacle-detection software. As a result of our collaboration, AUTO UNION Group achieved the outset development objectives:

23 weeks
First Version Delivery Time
3 weeks
Testing and Final Deploy Time
6 months
Overall Outstaffing Time
0 days
System Downtime after Final Launch
0%
Exceeding Budget
0 days
Exceeding Deadlines

CONTACT US

Let's talk about your project.
Use the form to drop us an e-mail.

Thank you for your message! Your request will be carefully researched by our experts. We will get in touch with you within one business day.