To autonomous driving with self-service

The simulation of autonomous driving functions generates large volumes of measurement data. To be able to process this efficiently, Trivadis has implemented a self-service portal in the AWS cloud for a German car manufacturer.

In Short

CHALLENGE

Parallel evaluations of exponentially increasing data volumes.

SOLUTION

Data management via self-service portal.

BENEFIT

Independent provision, processing and development of functions.

Our Solution

Automotive manufacturers have been working on implementing autonomous driving since the beginning of the last decade. One of the most important foundations for this is that the microprocessor systems, sensors and actuators installed in the vehicles work together in an effective manner. This is the only way to ensure that autonomous vehicles can move independently and safely.

Autonomous driving functions are therefore tested over thousands of kilometres driven both virtually and in reality. To release them onto the market, the measurement data must be processed efficiently and precisely. The prerequisite for this is that the required environments are provided flexibly and costefficiently.

Parallel evaluations of data volumes

A leading German automobile manufacturer generates several terrabytes of measurement data in its autonomous driving function tests. These will grow exponentially in the future. However, it was not possible to keep sufficient computing and storage capacity free on employees’ PCs or laptops for the parallel processing of this measurement data. For this reason, the required environments were to be provided via a self-service portal. Trivadis was tasked with creating a solution for this.

Targeted enhancement of autonomous driving functions is not possible without a solution like this. By using the portal we have developed, it will be possible to develop, secure and release highly automated driving functions for series production in the long term. This will make driving easier for lots of people.

Head of Vehicle Development

Cost-efficient and scalable application on a modern cloud platform

The portal was implemented on AWS Cloud. Various teams of partners, Trivadis and the customer worked together using a Scrum-of Scrum approach. Ideally, a Scrum team consists of three to nine people. For projects as large as this, however, the Scrum team is not any bigger, but instead several individual teams are formed to work together on the project. Very high security requirements also had to be taken into account. A self-service portal was developed for the vehicle developers, which can be used to load the corresponding measurement data into the preconfigured environments, process it and then store it again in a central memory. This portal enables all environments to be managed centrally. Even though the self-service portal is based on AWS, during development it was ensured that porting to other cloud providers was possible. 

Enhancing cost efficiency and security with self-service

This solution provides a cost-effective and scalable application on a modern cloud platform. The self-service portal fully automates the provisioning of the necessary environments, which means optimum support for the department – while also taking into account all security requirements.

Technologies used

Angular
Golang
Apache Guacomole
AWS Services (EC2, ECS, S3, ALB, Cloudwatch, …)

Prometheus/Grafana
Bereitstellung als IaC (Code)
EMR