There are 3 key vehicles in forestry that could be exploring autonomous features; Harvesters, Forwarders, and Haulers.

Harvesters, responsible for felling the tree, require a perception system that can identify the target tree. This requires computer vision models to be able to characterise a tree by size (age) and species. It could also help with characterising the optimal bucking length .These models will likely first hit the markets as a driver-assisted feature, whilst in the future it could be offer fully autonomous solutions. 

Forwarders are responsible for transporting the tree from the felled site to a road-side pick up spot or clearing. Autonomous forwarders could soon offer autonomous features, including aspects of tree detection, path detection and localisation. These features present a difficult challenge to achieve in the near future. However, there are solutions being developed in autonomous vehicle development that could enable perception, control and safety algorithms to safely navigate and execute autonomous forwarding. 

Haulers, or Logging Trucks, are responsible for taking the log from a roadside hub to the mill. These are heavy-goods trucks built to carry extreme loads on remote roads. These vehicles, such as Einride’s T-Log, are already under development or on the market. However, certain restrictions on load capacity and public road use limit the value of these vehicles. That said, the engineering feat in creating an autonomous or driverless haulage vehicle is impressive nonetheless – and will become the norm in coming decades. 

There are many benefits in autonomous forestry vehicles. Whilst some associate autonomy with replacing humans, forestry offers a means to improve the efficiency and safety of practices hand-in-hand with the skilled operators in this industry. Decision making is one of the core problems autonomous systems face, so skilled workers will likely continue to semi-operate machines out of the cabin. 


Efficiency

Autonomous vehicles can operate 24/7. In Sweden alone, there are millions of hectares of forest, and demand for our natural resources is only increasing. Autonomy offers a way to maintain operations around the clock. A research paper explored the benefits in efficiency that could be achieved in switching to autonomous or even semi-autonomous forestry operations. 

Safety

Taking the operator out of the cabin is one of the core benefits of autonomous vehicles, shared across automotive, mining, agriculture, defense and more. Better perception systems, comprising many sensors such as Lidar, RGB cameras and radar, can theoretically ‘see’ surroundings better than the human eye. This can prevent collisions with obstacles, other vehicles and even people. Furthermore, in dangerous or harsh conditions, it can be both safer and more comfortable for operators to not physically sit in the cabin. 

Design

Radically new features, such as autonomy, offer radically new design opportunities. Rethinking the vehicle entirely. The Einride T-Log is cabinless for example. Forestry vehicles are designed with precision surrounding weight distribution and balance on slopes – designs could be reconsidered if the operator is no longer in the vehicles. 

Sustainability

In industries such as mining, early studies have shown how autonomous features can reduce fuel consumption and tyre wear. Furthermore, improving localisation to specific paths can reduce the negative effects of soil compaction that are widely concerning heavy-vehicle usage in this domain. Through consistent, calculated operations, autonomy can reduce the impact the vehicle has on the wider environment. 

For the foreseeable future, the forestry domain will remain highly challenging for vehicles to operate entirely autonomously. Despite potentially fewer regulatory issues compared with the automotive industry, the environment that forestry vehicles operate requires further research and development. 

Driver-assist features, however, could further mature in the industry to bridge the gap from manual onsite, to entirely driverless. Perception algorithms indicating target trees, localising the vehicle, or simply alerting the driver to an obstacle could all help foresters in their day-to-day activities.

Machine Learning will continue to mature in other industries and transfer the developments into forestry in years to come. When you also consider processes such as simulation, and how they have served other industries in developing autonomous solutions, it is clear that forestry can apply some of these technologies to enable the advancement of autonomous forestry.

Simulation is a crucial part of autonomous vehicle development. But simulation has not advanced in off-road domains as much as it has on-road. To continue to accelerate autonomous solutions in industries such as forestry, it’s important to keep developing innovative solutions. Both software and hardware advancements have been developed over recent years in a bid to achieve ‘self driving cars’. Sensors such as Lidar scanners, ECUs, even cameras, have all advanced rapidly in the past decade to serve autonomous car development. These can be explored in off-road industries beyond the flagship ‘autonomous car’. 

At Repli5, we offer synthetic data and simulation environments to facilitate the development of autonomous solutions. 

Repli5 serves the forestry industry to help vehicle manufacturers develop their autonomous solutions. If you have an questions or would like to learn more about our solutions, please contact us.