The curse - and blessing - of boreal forest
by GIM Robotics
The current news feed from conflict zones is dominated by views of open fields, scattered forest patches and ruined urban settings. Navigation strategies for both crewed and uncrewed vehicles are naturally created to comply with the situation. Social media channels are filled with videos showing autonomous non-road vehicles moving seemingly freely towards distant goals avoiding solitude bushes of trees on open fields. Marketing people are talking about free navigation and expect all viewers to watch in awe as they are zigzagging around obstacles.
If you are new in mobile robotics, you might find it awesome and expect it to be created by some of the most advanced and novel AI solutions. Nothing could be further from the truth. There are many ways to get things done but widely used simplified potential field method, introduced by Professor Roland C. Arkin in late 1980’s, based on simple potential fields, representing behaviors (or schemas), of repulsion(avoid-obstacle) and attraction(move-to-goal), complimented with some local minima solving actions, like random potential fields, when summed (super positioned) provide that cool looking navigation.
Adding some uniform, perpendicular and tangential potential fields, the robot will excel in some more advanced navigation scenarios. As a very generic example, you could easily make a mission plan for your robot to move towards a distant hill over a flat open field with some sporadic trees and bushes, send the go command and enjoy the show. For that, you don’t even need GNSS availability, or any other GNSS-free global localization capability.
In normal boreal forest area settings, you usually see nothing other than numerous trees and bushes and often no clear path to proceed. In many places undergrowth is thick and can hide some dangers especially for the smaller robots and vehicles. These include for example some wet spots and larger rocks. For larger machines undergrowth is not a real issue, but big trees are blockers even for them.
In short, this environment turns a straightforward open field free demonstration into a challenging full-scale navigation exercise with mapping, localisation and motion planning. Some existing paths are man-made and mostly a priori known when operating on home grounds, others must be created on-the-fly onsite as the machine advances. The former enables planning ahead; the latter is the case where it gets interesting. This is the time when you need all your skills to enable successful navigation, especially if active sensors are not to be used for obvious reasons. The positive side is naturally the good protection that the forest in general, and canopy in particular, provide against various aerial and other threats.
That is what project BadB is all about in respect to its ground element, i.e., how to make your UGV navigate successfully also in our challenging conditions to achieve the mission objectives, whatever they might be. Based on our extensive experience, we at GIM Robotics know what we are doing and we trust in the benefits we will gain from the interoperability activities with our project partners.
The world calls our main mission area boreal forest. We Finns called it home.
