Mistra Digital Forest is paving the way for robotics in the woods
Under the research programme Mistra Digital Forest, the Swedish forestry industry is joining forces with the academic sector to create digital solutions for an efficient forest bioeconomy. Research at Umeå University is intended to give machines in the forest the capacity to perceive their surroundings and predict the effect of various decisions using digital physics and sensor data. The idea is that this will result in forest machines with smart decision support and autonomous functions.
Digital forest segmentation
Access to highly detailed data within the forestry sector is on the rise and is providing new possibilities in precision forestry. By interlinking the information flows in smart planning and decision support on different levels, the forest can be exploited more efficiently and with greater care, for the right products.
Mistra has therefore recently decided to fund the research programme Mistra Digital Forest, allocating SEK 58 million for the first four years. Skogsindustrierna (the Swedish Forest Industries Federation) is the host organisation for the programme, which is run by a consortium of universities, institutes and forestry companies. Part of the research will be conducted at Umeå University under the guidance of Martin Servin, senior lecturer at the Department of Physics and UMIT Research Lab.
"Our research will focus on giving machines in the forest the capacity to perceive their surroundings using digital physics and sensor data. When the machine interacts with its surroundings, new information is produced which can give new insights, if it is combined with models and other information," Martin Servin explains.
Today, it is difficult for both people and machines to interpret the information in their surroundings and predict the effect of various decisions and alternative ways to steer the forest machines. It is a matter of segmenting the surroundings into objects with different forms and physical properties; differentiating between stable and instable ground, between bush and stone, and planning how an unstructured pile of logs should be loaded.
"But in a virtual environment, driven using physics-based simulation, things are different. It allows us, under controlled conditions, to produce masses of synthetic sensor data with known connections and train models for perception of the surroundings, movement planning and decision support. We can also take into account the long-term effects of various decisions on the ecology. Examples included are reduced damage from driving, reduced emissions and the preservation of key biotopes. The smart decision support function is important for increased robotisation of the machines and decreased workload for the drivers shall contribute to better environmental values, not worse".
A joint testing site, Digital Forest Test Site, will be established as an important hub for collaboration in the research programme. Highly detailed data will be collected and field testing will be conducted with new solutions.
About Mistra Digital Forest
The research programme Mistra Digital Forest is funded via Mistra's call for proposals Bioeconomy and the forest and will run for a total eight years. The vision for the programme is to create digital solutions for a sustainable and efficient forest bioeconomy. The programme is led by Skogsindustrierna and the consortium consists of BillerudKorsnäs, Holmen, SCA, Stora Enso, Sveaskog, Södra, SLU, IVL Swedish Environmental Research Institute, Skogforsk, Umeå University and KTH.
Umeå University's part in the grant of SEK 58 million is SEK 8.4 million. Apart from research at UMIT, the Swedish Center for Digital Research is involved, led by professor Jonny Holmström at the Department of Informatics at Umeå University.