Cleanliness as a calling card. Cities want to be seen as an attractive place to live and work. Cleanliness has a major impact on the appearance of a location – and thus on urban development. Not surprisingly, considerable resources are invested in keeping cities from Paris to Tokyo clean. But when it comes to cleaning, there is a lot of room for improvement, as startup Cortexia from Western Switzerland discovered. Andréas von Kaenel, who co-founded Cortexia with André Droux, says: "Cleaning is often done where it’s already clean and not where it’s necessary." His partner, von Kaenel, adds that cleaning is mostly based on experience and habit, and not on actual pollution.
Cortexia gained this insight from a survey carried out by the company in ten European cities. It also revealed that some cities spend an above-average amount on cleaning, but the general public is still dissatisfied with the cleanliness.
Cortexia decided to take on this problem. The company wants to improve the cleanliness of cities and simultaneously reduce costs and environmental pollution. They are doing it with Machine Vision (MV) and Artificial Intelligence (AI). The Cortexia solution is already being successfully used in numerous European cities. It consists of a camera, an AI-enabled embedded computer from Syslogic, AI algorithms, and a web-based customer interface.
The hardware, consisting of a camera and an AI computer, is installed on vehicles that regularly travel in cities. That includes municipal vehicles, courier bicycles, postal tricycles or buses. The camera captures the surroundings. The images are not stored, but processed directly by the Cortexia Box. The level and type of pollution are determined in real time. Andréas von Kaenel says: "Right from the start it was clear to us that we wanted to carry out the image analysis on the Edge." According to von Kaenel it has two key advantages over the downstream evaluation. The data volume would be kept small and data protection would be ensured, as no images from the public space would be stored.
This is due to the fact that the collected images are processed and evaluated in real time in the vehicle. The vehicles can travel at up to 50 km/h (30 mph). All pollution types are assigned to categories. Depending on the category, measures for removal are suggested. Cities can also define pollution categories, be it syringes, broken glass or excrement, which they want to have removed immediately. If such contamination is detected by the system, an alert is triggered. Immediate removal can be ordered, so that a cleaning team can be sent to the appropriate location.
This example shows that Cortexia does not only assess the cleanliness of a city. Rather, the company provides significant support to cities in improving cleanliness and thus increase satisfaction of the local residents. Results from surveys also flow into the Cortexia solution. Pollution types that are perceived as particularly disturbing by the population can be prioritized for removal.
Cortexia can now use its system to measure exactly how successful this campaign is. Cortexia also measures the effectiveness of additional trash cans and ashtrays. Andréas von Kaenel says: "Cities have a lot of tools to get cleanliness under control." So far, however, they have not had the opportunity to measure the effectiveness of individual methods.
Using Cortexia’s web-based customer interface, city administrators can pull up the current condition on visualized maps in real time. It shows which streets or neighborhoods are heavily polluted and which areas are too clean. Too clean? Yes! In places that are too clean, resources were invested in cleaning that could be better used elsewhere. Suitably, Cortexia is able to show the cities how the situation can be improved by pointing out how existing resources can be used as effectively as possible, where savings potentials exists, or where additional resources would make sense. Cortexia significantly increases the effectiveness of city cleaning. For every dollar that cities invest in monitoring at Cortexia, five dollars can be saved on cleaning. This average has been derived from evaluating previous mandates. The increased effectiveness also reduces environmental pollution. Unnecessary trips by cleaning vehicles are prevented, thus reducing emission. Ineffective cleaning procedures are uncovered and can be prevented in the future.
During the test phase, Cortexia had designed the first prototypes of the hardware with the help of an engineering office. For the serial production, Cortexia looked for a suitable hardware supplier. Embedded specialist Syslogic won the bid.
André Droux, who is in charge of product development at Cortexia with a team of engineers and computer scientists, says: "We were looking for a company that could supply us with robust, compact, AI-capable computers." According to Droux, Syslogic impressed them with its uncompromising industrial embedded computers during the evaluation and was able to demonstrate well-founded in-house expertise with AI computing. With Syslogic, Cortexia has not only gained a hardware supplier, but even more so a development partner.
Within a few months, Syslogic developed a completely new device in close cooperation with Cortexia. Syslogic is one of the few European companies that develops and manufactures embedded systems in-house. The company has already gained plenty of experience with AI-enabled embedded computers in previous projects. This experience has been incorporated into the development of the Cortexia device. A Jetson Xavier module from the AI pioneer Nvidia serves as the processor platform. For Cortexia, Syslogic has brought a completely new board and housing design to reality. The AI computer has two PoE-capable (Power over Ethernet) LAN interfaces. It allows cameras to be connected to the device without additional power supply. Syslogic also integrates LTE, WiFi, and GPS features in the Cortexia device.
Specific focus went into the robustness of the AI computers. André Droux explains: "Our solution is used worldwide. The hardware is sometimes mounted on municipal vehicles, sometimes on bicycles or motorcycles." Accordingly, the AI computers must function reliably under very different conditions, says Droux.
These requirements pose no problem for Syslogic. The company has thirty years of experience with embedded computers used in or on vehicles. Syslogic supplies numerous manufacturers of railway vehicles, construction machinery, or AGVs (Automated Guided Vehicles). Droux says: "There are many AI-enabled computers now, but Syslogic was the only supplier that could deliver ultra-rugged AI computers for vehicle use."
The standout features of the industrial design include a robust housing with IP67 protection as well as an ultra-rugged electronics design.
Syslogic does not use any moving parts such as easily breakable fans. The AI computer is passively cooled thanks to a sophisticated housing design. SSD cards (Solid State Drive) from Cactus Technologies are used as storage media. They are among the world’s most robust storage solutions.
Cortexia has already received orders from several European cities, including Athens, Asti, Paris, Lyon, Toulouse, Zurich, and Geneva. Cortexia has also entered into partnerships with postal operators, since postal vehicles are ideally suited for equipping with the Cortexia Box. Another milestone in the still young history of the company is the French IoT Award in the Smart City category. Thanks to this award, Cortexia was invited to the CES trade fair in Las Vegas to present its solution.
Andréas von Kaenel says: "The interest in the USA was very high and many visitors already suggested new application areas for our solution, for example assessing the condition of road markings." These are quite interesting inputs, adds von Kaenel, which Cortexia will certainly keep in mind. But for now they want to focus on the cleanliness measurement and improvement and become the global leader in this field, says von Kaenel.
Hardware partner Syslogic also believes in Cortexia’s big plans. Florian Egger, who heads sales at Syslogic, says: "It’s incredible to see how much Cortexia has already achieved in a very short time. The collaboration between Cortexia and Syslogic is also very dynamic and has already yielded results within a very short time." As can be expected, everyone at Syslogic is excited that they can contribute to clean cities with their AI computers. Egger says: "Cortexia shows how cost-effectiveness and environmental compatibility can be improved by AI."
Measuring cleanliness in cities is not a new concept. New York was the first to do it in the seventies. In Europe, Zurich plays a pioneering role in cleanliness measurement. Zurich currently carries out 15,000 audits per year. The auditors travel with eBikes and record the current condition. In contrast to many other cities, Zurich has a cleanliness index that does not only measure cleanliness or pollution. All contaminations are divided into categories, which are also linked to their origin and include measures to remove them. This holistic approach has also convinced Cortexia. Accordingly, the company has chosen Zurich’s Cleanliness Index as the basis for its solution. In turn, Cortexia’s idea of measuring cleanliness with the aid of Machine Vision and AI was met with great interest by the City of Zurich. Expectedly, the City of Zurich and Cortexia maintain a close partnership.
Cortexia was founded in 2016 by André Droux and Andréas von Kaenel. The company is based in Châtel-St-Denis in western Switzerland. Currently, the company already has 10 employees. Financing for the coming years is largely secured and the opening of a French branch is imminent. In addition, Cortexia has already won numerous innovation prizes, including the French IoT Award, the Freiburg Cleantech Innovation Award, and the CE DeepTech4Good Award.