Increasing the automation of forestry machines can improve their performance and make the forestry operations carried out by the machines more sustainable. Therefore, this work package has made it its mission to pave the way for partially automated forestry machines.
The work includes analysing new machine systems that can better meet the requirements of economically and environmentally sustainable forestry. Models for object detection and motion control are trained using large amounts of collected and simulated data. This work package also explores how an increased level of automation can improve the machine operator’s working day, and simplify complex tasks.
The work package focuses on three tasks:
The potential of machine systems in the digital forestry of the future
Harvesting, thinning and other forestry operations have been carried out in much the same way for decades. However, the increasing availability of detailed data and the demands of addressing various societal challenges are now driving change. This task explores the potential of new machine systems to contribute to the digital forestry of tomorrow. Which technologies are worth investing in? By seeking answers to these kinds of questions we can avoid misguided investments, and we can accelerate the development of a digitalised forestry that contributes to reduced emissions, lower environmental impact and increased efficiency.
Enhancing the forest machine operator’s work environment
Forest machine operators play a key role in improving operational work in the forest. For example, having good prerequisites for doing their work can contribute to increased productivity, with reduced fuel consumption and lower emissions as a result. Other benefits of well-executed forestry operations are increased timber value and ensuring that natural and cultural-historical values are taken into consideration. Therefore, this task develops automatic support systems for operators, by providing direct feedback, for example. It also explores the possibility of online training for both new and experienced drivers.
Data, models and software for increased automation
This task aims to automate forestry using AI, robotics and simulation. Data from machines and equipment in the field are used to create annotated datasets. In turn, these are used to train models and algorithms for automatic control of forestry machines. This task tackles some of the obstacles on the road to increased automation: how can the process of collecting the necessary data be automated? How are models to identify forest objects best trained? And how can the development be accelerated using simulation? It also develops self-driving forwarders that operate safely, efficiently and are gentle on the environment. The results are disseminated outside the programme, for instance so that experts in robotics can contribute to the field.