Vision and navigation software: the technological layer behind autonomous systems

Everyone knows Google Street View. But few people know that there are companies that build such systems and then convert the images into usable data. Within the Autonomous Green Management cluster, Horus is one such party: a Groningen company that develops software and hardware to accurately map the environment.

That technology is an important building block for autonomous systems. Because vehicles that need to navigate autonomously must first be able to see and understand their surroundings. "We make technology to map the world," says Dirk Aalbers, Managing Director. "But for us it's not just about the pictures. It's about what those pictures tell you."

Horus develops software and hardware for mobile mapping: vehicles that capture the environment while driving with cameras, sensors and GPS. Those images are valuable in themselves. But the real question is what happens to them afterwards. "A camera sees a lot," Aalbers explains. "But ultimately you want to know: how many traffic signs are there, what is the state of the asphalt, what needs to be replaced?"

Horus builds the software that translates visual data into actionable information. This is done with a platform that adapts to what a customer already has. Whether it's off-the-shelf systems or customization, the technology connects to existing hardware and systems. "We sit between the hardware and the software," says Aalbers. "So that our customers don't have to worry about what's under the hood."

One of the key developments Horus is working on is moving data processing from the cloud to the edge: directly on the device while the vehicle is driving. "People increasingly understand that sometimes you want to run something that needs to be faster or cheaper than computing something in the cloud," Aalbers said.

By running AI models directly on the GPU of the moving vehicle, systems can analyze in real time without delay. That feature also makes Horus directly usable by partners in the cluster, such as KOWW. Horus' software can be placed on an autonomous rake boat, where it combines sensor information from GPS, camera and lidar with geographic polygons previously captured by drones. That way, the boat knows exactly where it is and which plants need to be removed. The calculations are done energy-efficiently enough to run on a battery and fast enough to make real-time decisions.

Through the Autonomous Green Management cluster, Horus came into contact with new partners, including Vector Machines. Together, knowledge is being exchanged around a scalable series of the systems, because this type of technology only really comes into its own when it is deployed on a larger scale.

"Networking in the local region is very valuable to us," says Aalbers. "But what also strikes me: there is a remarkable amount of knowledge and expertise in the field of geotechnology here in Groningen. This is the geo-silicon valley of the north."

In addition, Aalbers notes that the underlying principles of positioning, sensor integration and autonomous navigation are largely the same: whether mowing or water rooting. "You have to keep that knowledge together and learn from each other," says Aalbers. "The cluster makes sure that actually happens."

Wondering what else is happening in the Autonomous Green Management cluster?