Automation for the Forestry of the Future
SweTree is a Swedish biotechnology company focused on developing sustainable plants for forestry. In a comprehensive collaboration with Prevas, the company is developing an automated process for selecting and propagating the most viable trees already at the embryo stage.
For this project, Prevas has developed an advanced vision system where AI-based image analysis and robotics work together with extremely high precision.
- Our goal is to plant trees that deliver higher growth and better quality, says Magnus Hertzberg, CTO at SweTree.
- For society, this is about creating healthy, productive forests that can provide biomass while also contributing valuable ecological benefits, Hertzberg continues.
The method SweTree uses is called somatic embryogenesis, where new plants are created from plant cells in a laboratory environment. Traditionally, forest seedlings are produced from selected trees in seed orchards. It is an established method that works on a large scale, but with limited opportunities to fully benefit from breeding improvements.
- With somatic embryogenesis, we can select the strongest trees already at the embryo stage and reproduce them. This can result in trees that grow up to 30 percent better and are also more resistant to disease, insects, and drought, says Magnus Hertzberg.
Identification, Picking, and Planting – in Under Three Seconds
SweTree’s automation process consists of several stages, with Prevas contributing to one of the most critical: using a robot to pick embryos and place them into a cultivation medium. The process begins with a camera inspecting a large number of embryos. They are extremely small, typically between 1 and 5 millimeters long depending on the species.
- Using AI-based image analysis, we identify a suitable embryo and calculate a number of parameters, including position and angle. This enables the robot to pick up the embryo correctly, both to avoid damaging it and to ensure accurate placement later in the process, says Johan Sandberg, Vision Engineer at Prevas.
The embryo is then planted in a cultivation unit within a tray containing a total of 160 units. The openings in the cultivation units are also identified using AI, and the entire process is completed in under three seconds.
Delivering a Complex End-to-End System
- It is not efficient for us to build cutting-edge expertise in AI and robotics internally, so we collaborate with specialists. Together with Prevas, we can offer a complete robot- and AI-based system to customers around the world, says Magnus Hertzberg.
To realize the project, Prevas has contributed expertise across several areas, including machine vision, AI, robotics, PLC systems, and UX.
- This is an extremely complex project where we truly leverage our capabilities as a full-service provider. The collaboration with SweTree is a great example of what we can achieve in the more advanced part of the spectrum, even though we also work with simpler solutions, says Milad Abdhagh, Account Manager at Prevas.
