Welcome to the world of TPLE!

How a method for knowledge management of complex products emerged from a Corona-related long-distance relationship

November 2024: in this blog I would like to give you background information on the Typebased Product Line Engineering (TPLE) method created by Klaus Anwender and myself. After describing the status quo and explaining the benefits of TPLE in the first blog posts, I would like to write something about "how it all started" in this post.

Chris Seiler

My name is Chris Seiler , I am a software engineer at Mercedes-Benz Research & Development in Sindelfingen / Germany and my motivation is to be involved in the realization of the software-defined vehicle (SDV). In my view, complete mastery of the software domain is the key to a successful future for the German automotive industry. It is important to me that our children and their children can grow up in an environment and society worth living in. Of course, this includes not only ecological aspects but also economic ones.

In order to make the topic of software manageable in the automotive industry, fundamental changes in the "way of working" in the development departments of the OEMs and suppliers are required. Basically, one question needs to be answered: how do I deal with the growing complexity? We then quickly come to the topic of data management - there are some blueprints that talk about a "data centric enterprise" or a "data driven architecture". This is exactly the right approach!

Everyday development in the automotive industry (symbolic visualization)

What does this mean in the everyday lives of our fellow developers? What does this mean for the other departments, such as marketing/sales, product management, purchasing, production, after-sales service...?

We have an answer for that, and it's TPLE. I hope this gives you an idea of what to expect in the next blog posts.

How did TPLE come about?

Thorsten Legat

I still remember exactly how I heard Klaus for the first time in some MS Teams sessions at the beginning of 2021 with a lot of participants (I didn't see him for a long time because he always had his camera switched off). He always seemed a little arrogant to me because he asked specific and uncomfortable questions from the perspective of a system engineer. He also had a profile picture that always reminded me of Thorsten Legat, a former german soccer player. But over time, I started to pay attention as I listened more closely to Klaus. And this encounter with Klaus changed my life.

Klaus Nutzer

He had written a research paper entitled: "Product Line Engineering - Feature Management (object-oriented and feature-based)" and wanted to submit it for the final examination to become a "Certified Systems Engineer (GfSE)" ® Level A. Unfortunately, his work was not recognized and so he was unable to complete his degree.

Since we had already developed our own approaches to reusing knowledge fragments in MB.OS at the same time, a number of things from Klaus' work seemed very familiar to me. And since then we have rounded off the topic together, completed it, questioned it, promoted it, cleaned it up one by one. And I still work in this mode.

We brought TPLE as input from Mercedes-Benz into the SofDCar research project, which ran from mid-2021 to November 2024. In addition to mastering product variance, one focus is the fully digital homologation process for purely software-based feature upgrades. This enables the implementation of continuous homologation - the associated white paper was created as part of the digital.auto initiative and can be downloaded here .

Klaus has been in passive partial retirement since autumn 2023, but he is still "The TPLE Brain" and is available as a TPLE expert through the company UserKnowHow - click here for the contact form .

During the entire Corona pandemic, we communicated almost daily via MS Teams and together developed the TPLE concept and brought it to life.

Thanks to Startup Autobahn I discovered SPREAD

The Startup Autobahn Expo was in July 2021. Due to the Corona pandemic, this event only took place online. I became aware of the project How a Product Intelligence Platform Accelerates a Cross-Functional Digitalization Initiative by Creating an Exact Representation of All Engineered Products - why? Because SPREAD.ai 's approach - bringing the knowledge of engineers into a graph-based data model and thus creating a semantic structure - is the same as TPLE. The specific project was about bringing together the different data sources such as design data of the wiring harness with 3D data and diagnostic data in order to be able to prepare the possible causes for people very quickly in the event of errors in production.

What fascinated me was the realization that the wiring harness in the E/E networking world has a similar role to variant coding in the software world: both integrate generically built components (E/E: the control unit hardware, software: the code blocks of the control unit software) into the individual context of the currently built vehicle.

A screenshot from the TPLE Manager developed by SPREAD.ai

We started the first prototype implementation of TPLE in April 2022 together with SPREAD.ai . However, we quickly saw that frequent changes to the data model meant that the developers very quickly reached the limits of the classically programmed front-end application. During this time, however, we learned a lot and constantly questioned and further developed the TPLE concept. The key players from SPREAD.ai in this project were Philipp Noll , Daniel Metzinger , Daniel Wilms and Frederik Günther . We were supported at the time by Valerij Asmus and Marcel Engelmann from Mercedes-Benz Management Consulting.

In parallel, NTT Data consultants supported us in assessing the maturity level of TPLE - key players were Leonid Borodulkin , Matthias Fickler , and David Pastor Sampedro .

In December 2022, I met Dirk Slama on the initiative of Achim Nonnenmacher from Bosch. This encounter also changed my life forever. Dirk is the brains and driving force behind the Bosch Connected World conference, initiator and brains behind the digital.auto initiative. At the same time, he is a professor at the Ferdinand Steinbeis Institute in Heilbronn. I will write how the story with Dirk and digital.auto continues in a separate blog post. He is now the most important supporter of the TPLE idea that I have been able to find so far.

Here I explain to Christian Heissenberger and the students during the hackathon what is behind TPLE

Back to the cooperation with spread: with the status of the spread-based TPLE Manager, we announced the hack challenge "Welcome Scenario" in November 2023 - I wrote my experience report on LinkedIn in this article about the SofDCar Hackathon .

This is how Christian Neusius helped us out of a jam

At the end of 2022, Klaus and I realized that we had to make a complete restart - starting with a blank sheet of paper, we had to set up the technical implementation of TPLE cleanly from scratch. Due to previous collaboration, Klaus and Christian Neusius knew each other very well, and I met Christian in February 2023. Again, this encounter changed my life forever. In fact, the two were able to virtually lock themselves in for 5 weeks in the summer of 2023 and develop the TPLE data model cleanly from scratch.

And within a few weeks, Christian Neusius was able to use his code generator to build a cloud-native app including a graph database and GraphQL API in the backend and a Typescript-based web frontend -> the TPLE Manager Reloaded was born.

Visualization of the Feature Data Model in TPLE Manager Reloaded

When changes or extensions are made to the data model, we are now able to regenerate all the artifacts derived from it at the push of a button. This gives us several advantages: very high speed and therefore very high flexibility, while at the same time offering very good quality at a reasonable price, because manual work is almost no longer required in this workflow.

We shape the world of TPLE

Klaus and I have been working together on the topic of TPLE for almost four years now and we have come a long way. The SofDCar research project is in its final phase and the question arises: Quo vadis TPLE?

I see two storylines here:

1. Bringing TPLE into series application In my view, SDV can only work if we use TPLE across the industry. To do this, the first step we need to take is to place the TPLE data model as a new open source project, for example in Eclipse SDV. In addition, we need concrete use cases with concrete added value so that support for a TPLE-based future can grow.

   
   

2. Answering open research questions in further research projects For example, it is not yet clear how we can move from today's set- and rule-based variant management logic to a target image in which TPLE realizes the central variant management for software and the hardware-based variance can be intelligently integrated using code rules.

What do you think about the future of TPLE? What benefits could you get in your context if you had TPLE at your disposal? Feel free to leave a comment or send me a message. Stay tuned for more to come soon!

And don’t forget: keep on rocking in a free world!