Agile working at Autonomous Intelligent Driving

Share this

Completely autonomous driving in urban areas is the mission of Autonomous Intelligent Driving. With agile thinking, the Audi subsidiary is creating the space they need to make this vision a reality by the beginning of the next decade. Insights into a working process that produces much more than just brightly colored Post-it notes.

9 a.m. on Wednesday morning – a cheerful “Daily stand up, you guys” can be heard throughout the open-plan office in the north of Munich. Nastaran Matthes stands in front of a wall full of colorful Post-its, a red ball in her hand. The instant the Cloud Service team is assembled, it starts: What are they working on right now? What problems are they facing? How can they be solved? The red ball is passed from team member to team member. Less than ten minutes later, they have a plan for the next 24 hours. Nastaran is an agile coach at Autonomous Intelligent Driving (AID). That means that she plays a lead role in developing an agile company culture.


The customer is king

As a center of expertise for autonomous driving in urban areas, AID is developing a software module which, in the future, will guide cars through the city without human drivers. With a startup mentality, the Audi subsidiary is creating the space for itself to think and act differently. Because developing software demands speed and flexibility, especially when you’re developing something that sounds like science fiction right now. And that’s where Scrum comes into play. The agile working process makes it possible to react quickly to changes, involves customers in the development process, and allows employees to assume more responsibility. “Only when we continually take customer needs and market changes into account can we ultimately develop the right product,” Nastaran explains.

Like all teams at the Audi subsidiary, the Cloud Service team uses Scrum methodology. It is developing a cloud infrastructure that is used to collect data. Software engineers then use this data to train the neural network – the central building block of the “brain” of the car’s artificial intelligence..

The Cloud Service team is currently in the midst of a sprint, which is a two-week long development cycle. Each cycle produces a fully integrated, functioning software that is then implemented into the car’s software stack. Every morning, the team meets to discuss the next tasks; other than that, meetings are avoided as much as possible.
More trust, less control
The motto is “more trust, less control.” From planning, through implementation, and right up to the demo for the product management team, all team members are equals. The result: “People who assume responsibility take more pleasure in their work and show greater dedication,” says Nastaran.
Sami Vaaraniemi is tech lead in the Cloud Service team. He bears technical responsibility for the sub-product. “The most important quality attribute of software is its ability to support change. The main goal of our tests is – in addition to the correctness of the software code of course – to make the code base changeable. This allows us to continuously update individual pieces of code without introducing regression bugs”, says Sami. In order to achieve this, he and his team develop iteratively, meaning that they deliver the software step by step. This allows errors to be recognized more quickly, and the problems can be dealt with immediately. Nastaran assists the team in an advisory capacity during the entire process: “My job is to overcome obstacles and make sure that the team doesn’t lose sight of their goals.”
A must-have: an agile way of thinking

Scrum is popular because it requires less roles within a project. But it is only one of many ways to implement agile working processes. Nastran explains that what really matters is the thought processes and mentality on which agile working methods are based: “The driving factors behind an agile mindset are motivated employees, autonomously organized teams, human interaction, and continuous improvement.” The premise of continuous improvement applies to both the quality of the software and to the process of working together. This is why each sprint ends with a retrospective—a review, in other words. “It isn’t about placing blame, but about searching for causes. We’re looking for the “why,” not the “who,” says Nastaran. “Everything that we can improve on goes into the to-do list for the next sprint.” It’s relatively easy to work with agile principles in a small team. That’s why startups were quick to adopt these working methods. But in order for the individual software and hardware components to ultimately fit together in a car, all teams need to be synchronized. “The great challenge there is to stay lean, without too much bureaucracy,” says Nastaran. “But if we can get the agile culture to really take root now, then later we’ll only need to water it.” And that’s exactly how the AID wants to make the dream of robot taxis a reality in the near future.