Pallon – Artificial Intelligence for urban drainage systems

Public sewers worldwide are aging and are thus more prone to defects such as cracks or corrosion. ETH spin-off Pallon (previously Hades) uses AI to detect these defects in sewer inspection videos automatically.

The team’s vision is to improve the management of sewer systems and thereby make cities more sustainable and resilient. An interview with Christian Koch, co-founder of Pallon.

Christian, what attracted you to the rather unpleasant urban drainage systems?

Dominik, one of the co-founders, and I are both environmental engineers and very familiar with urban drainage systems such as sewers. When discussing our plans for the future, we remembered one of our fellow students complaining that he had to watch hours and hours of sewer inspection videos during his internship at an engineering firm. It was tedious and he definitely did not want to do this kind of work ever again. That is how the idea was born.

We then took a closer look at the subject by reading articles and conducting expert interviews. According to scientific literature, about 25% of the defects are currently missed or misclassified. In addition, the defect reports and the resulting sewer condition assessment are also very inconsistent, as every expert records different defects, resulting in a high level of uncertainty. Therefore, 15-20% of sewer pipes are replaced too early wasting money, or left in too long polluting the environment. Lastly, the data is often stored locally in an unstructured and not easily accessible format. This limits collaboration between stakeholders and hinders the use of advanced data analytics to support decision-making.

picture of the inside of a damaged drainpipe
Sample image from a sewer inspection showing a defective connection on the upper left side of the image as well as several cracks.

What are the benefits of Pallon’s technology?

Instead of watching videos and manually reporting defects, our customers only need to upload their video to our cloud. There, defects are automatically detected and classified using computer vision and deep learning. Outsourcing the task to a machine is not only much faster but also produces objective, reproducible and comparable data. The data is automatically stored in a secure cloud environment from where reports can be generated or data can be integrated into the geographic information system. Overall, this creates the basis for a data-driven infrastructure management of sewers.

Who are your customers, and what are their needs?

Our main customers are public sewer asset managers such as city councils and sewer service providers (e.g. inspection companies and engineering firms). We are currently testing our technology with a few selected city councils and service providers in the DACH*1 region and are continuously improving our services on the basis of their feedback. Working with both sewer asset managers and service providers early on, helped us to understand our customer’s requirements better. While sewer asset managers tend to be more interested in higher data quality, service providers use our technology to increase their efficiency and to be able to analyse more sewers with the same number of employees. Both stated that they have difficulties recruiting skilled workforce. Thanks to our services, they can overcome this challenge.

What challenges did you face?

Initially, it was difficult to recruit talented computer scientists experienced in deep learning as they are currently in high demand. Being at ETH definitely helped us finding skilled computer scientists and enabled us to collaborate with research labs.

Then, of course, it was important to secure the necessary funds to develop our technology. We were very lucky to be supported by several institutions such as BRIDGE proof-of-concept, EXIST Gründerstipendium, Climate-KIC or Swiss Climate Foundation, which allowed us to overcome our initial funding gap.

What are your goals for 2020?

In 2020, we plan to grow our customer base and we are looking for interested users and strategic partners. Furthermore, we will continue to expand our team and will be looking for new talent. Lastly, we want to launch our first investment round next summer. We would be excited to get in touch with investors who share our vision of making cities more sustainable and resilient.

team photo Hades - 4 men
The Hades co-founders: Christian Koch, Eric Wolf, Anton Middelhaufe and Dominik Boller (left to right).

Contact / Links:

external pagePallon

Do you want to get more "News for Industry" stories?

external pageSubscribe to our newsletter

external pageFollow us on LinkedIn

Are you looking for research partners at ETH Zurich?

Contact ETH Industry Relations

JavaScript has been disabled in your browser