Unveiling Drainiac.ai

5 min readOct 24, 2021

We’re happy to unveil our project progress. With Drainiac.ai we iterate towards AI workflow automation services for the sewer industry. Our first products are APIs and Portals which aim to assist and ultimately fully automate sewer system CCTV survey video analysis by creating the “world’s most experienced AI inspector” © (patent field EP21216170). To achieve this we’ve been developing under the hood a robust and scalable backend architecture to potentially automate more use cases from additional industry verticals.

Challenge

For example in Germany ~0.6 Mio km public and ~1.8 Mio km private sewer systems require maintenance. It’s estimated that in some regions ~80% of private sewers systems are at least partially broken. Sanitation and keeping our cities’ sewer systems in a good condition are important foundations especially for our urban populations. Bill Gates tackles these challenges on a different level. An essential part of the sewer system maintenance are CCTV video surveys. Here we aim to help with Drainiac.ai and our workflow automation solutions.

Looking back

We began the journey by asking ourselves how we could utilize our computer vision and deep learning know-how gained in the autonomous driving and robotics field to solve challenging but less mainstream problems with intelligent software solutions.

Then we started thinking more concretely beginning of 2020 with my friend Dave. For example, we tried to build a stereo camera perception module for autonomous commercial lawnmowers based on the ZED and Jetson Nano (link to prototype video). Triggered through my friend Nick we took a closer look at sewer systems and the dynamics of the industry. When relatives received a CD with a handful of seemingly manually re-named images of a CCTV sewer video inspection they requested we got the feeling that there might be a good opportunity for digitization and automation. After talks with a few experts, researchers, operators, and contractors we got a better understanding of the potential, collected first datasets, and started drafting a go-to-market plan. The sewer and water infrastructure industry in the US and Europe alone is a multi-billion dollar market opportunity.

Product

Our SaaS product is completely hardware independent, fast, objective, accurate and relevant. It helps everyone who wants to save time and/or add a quality check to CCTV reports. It also holds the potential to change the complete survey industry where operators today spend hours on site annotating defects. In the future we believe that with a product like Drainiac.ai operators can do more surveys a day and the video analysis itself will be completely automated. Looking forward to demonstrating our products and feel free to book a demo with us.

Viewline (UK)

It was luck and good timing that the team from Viewline was interested in AI and it’s sewer applications when we approached them last year. We started our collaboration in September 2020 and findings were concluded last month (link to our announcement video from last year).

t4 Spatial (USA)

In our project with t4 Spatial together we aim to add more AI enabled features to the t4 Vault. It has been a real pleasure kick-starting our collaboration with Mike and Rogelio. We are more than happy to see already first results of our joint efforts materializing with the goal to simplify & speedup customer workflows even further.

Looking forward

Our long-term idea with Drainiac.ai is to create the “world’s most experienced AI inspector” © for CCTV survey video analysis catering towards all major classification standards (cf. Waymo’s most experienced driver).

Learning opportunity

Working full-time plus overtime for the BMW Group and starting Drainiac.ai on the side is a challenge but worth it. It’s a valuable opportunity and combination for faster learning. At BMW I support in my role as project lead our goal to deliver high-quality software with a premium user experience at a global scale with potentially millions of users in a relatively defined setting. With Drainiac.ai we can be creative about everything we do. From zero to commercial launch. Why, how, what. Technologies we apply, partnerships we aim to build, product architecture, features, prioritization, working style, business model, strategy, target users and more. Sharpening an entrepreneurial mindset. Both from a product development point of view as well as from a leadership perspective. Learning from mistakes in this setting feels also easier for me because the scale and effect are much smaller.

Drainiac.ai has been and continues to be an interesting learning opportunity: Where before for me it was an onboard robotics focus in a research setting relying mostly on ROS, C++, OpenCV and TensorFlow (e.g. ShadowCam). It’s now about a robust and scalable backend application in a production setting where we work with Python, FastAPI/REST, Postgres, Rust, Go, Nginx, Traefik, Docker, RabbitMQ/AMQP, React, Nest and more.

Acknowledgements

On a more personal note I would like to take the opportunity to deeply thank my project partner and friend Magnus. We met at CDTM’s Class of Fall 2015. Magnus is essential to Drainiac.ai and only with him, his dedication and experience we were able to build our product so well and so quickly. Our thanks also goes to Shahzil who has joined our team recently and is the force behind our Drainiac.ai Portal. We would like to thank Mindinventory for their support during ramp up.

I would like to thank Prof. Daniela Rus, Heidi Wyle, Dave and the whole DRL community. Being allowed to learn through building autonomous cars & robots together at MIT and Venti Technologies (Venti recently announced a $8 Mio. funding round) was an amazing opportunity I am very grateful for. Now we are essentially bringing autonomous driving technology to sewer systems.

We want to thank Michael Fischer from Venner Shipley LLP for his tremendous support during our patent filing & application process (EP21216170) and are grateful for the funding from WIPANO.

We want to thank Messe Stuttgart and VDMA for being recognized as finalists of the Vision Startup Award 2020. Thomas Becker from GOF has been a strong force helping us to create sound legal frameworks. We want to thank Amazon Web Services (AWS) and Dennis for their support through the activate founders program and our friends Stefan & Marvin from STACKIT for providing us a secure and reliable server infrastructure made in Germany.

Last but not least, we want to thank our mentors and advisors who support us from time to time and help us with their experience: Prof. Dr. Brügge (TUM), Prof. Dr. Regensburger (THI), Dr. Möller (CMO at PEIQ), Dr. Franz (CEO at Framos), Dr. Göpfert (CEO at ID-Consult) & Dr. Stetter (CEO at ITQ).