“I would say that this solution has filled a blank spot in the industry with its affordability and simplicity. The robots are built on ai-technologies with the production in mind. It requires very little programming to adjust their routes, and it is possible for colleagues in the production to program them themselves or manually drive the transporters. This has not been seen before,” says Per-Lage Götvall, senior researcher at Volvo Group.
Tackling a Growing Logistics Puzzle
As Volvo Group introduce new propulsion technologies into truck production, the production assembly lines are populated with a more diverse range of models than before. To accommodate this increased complexity, Volvo Group has implemented a mixed model assembly concept, enabling flexible production having the different truck models manufactured on the same main line. This presented a logistics challenge, as a growing number of unique parts needs to be delivered to the main assembly line at the right place and the right time to be assembled on the truck.
Traditional methods, like manual industrial trucks and part trains, would quickly reach their limits. “The factory was already crowded,” Per-Lage explains. “And existing automated solutions didn’t offer the scalability or flexibility we needed. It forced us to think completely differently.”
The pondering with another perspective led to a relatively simple idea: what if the robots could “see” the factory from above? By mounting cameras in the factory ceiling and using relatively simple QR codes for navigation, the team found a way to give transporters a bird’s-eye view of their environment to be able to navigate. “It is an invention that is advanced, yet surprisingly simple,” says Per-Lage Götvall.
At the heart of this innovation lies the Generic Photo-Based Sensor System (GPSS). This advanced system uses ceiling-mounted cameras, deep learning, and computer vision algorithms to detect and track humans, robots, and other traffic. Unlike traditional autonomous robots, these transporters are very simple and rely on the external navigation system, making them simpler to maintain, and easier to use for the end-user.
From Serving Coffee to Transforming Production
The journey to bring the solution to life began humbly, with the research team experimenting with robotic lawnmowers and involving students from universities like Chalmers University, Örebro University, and EPFL in Switzerland. The solution developed well, and the potential of it all started to become noticeable. Their first prototype of the system had a simple but memorable debut: serving coffee and snacks at the inauguration of CampX, Volvo Group’s global innovation arena site in Lundby, Gothenburg, including several prominent guests.
From there, the project took off with new speed with the goal of entering production. The transporters were put to the test in the Tuve pilot plant in Gothenburg, a production simulating area made for testing new technologies and introducing new products. And additional researchers were brought onboard, among others Kristofer Bengtsson, senior researcher in smart and connected operations at Volvo Group.
“We could see quite quickly that this solution worked better than expected, so the decision to try it out in a full production was easy to make,” he says.
Unified environment
The GPSS solution has really been put to the test from many aspects, not only it’s technical functionalities. Safety is one of the absolute key values of Volvo and Safety and the robots also needed to be ready to work among both human colleagues and other technologies without any risk of complications.
“With the rigorous testing we have done at the Tuve site, we ensured that they met our high standards when it comes to safety and integrity of our colleagues. We have actually created a system where humans and robots are working safe and effectively side by side, in what we call a unified environment. The focus on combining both automation and human flexibility enables a very robust system.
“A factory will be able to install the cameras and r robots, fine-tune the AI algorithms, and adapt the system by themselves,” Kristofer explains. “This makes automation an integral part of our daily improvements, not a one-time investment.” This is possible due to a combination of factors. One example is that everyone in the factory can drive the robots manually if they are blocked by something. Other factors are the simplicity to change how it is moving and its behavior yet keeping the system safe.
The intelligence behind the machines: GPSS
The core of this innovation is the Generic Photo-Based Sensor System (GPSS), which uses ceiling-mounted cameras, deep learning, and computer vision to track humans, robots, and other traffic. Camera images are processed in a Kubernetes cluster using GPUs for fast semantic segmentation, identifying obstacles with minimal latency. The results are merged into a unified map used by the control system, which sends navigation commands to robots over Wi-Fi. Most of the system, including AI inference and the robots' embedded controller (ESP32-C3), is implemented in the programming language Rust.
The Future of Factory Logistics
The success of this transformative solution at the Tuve plant in Gothenburg has paved the way for its introduction at other locations within the Volvo Group. Plans are underway to implement this cutting-edge technology at other major truck production sites, ensuring a brighter and more efficient future for the industry.