PhD · NTNU · path planning
Planning a long way ahead, around obstacles
An RRT*-based planner that looks beyond the next waypoint: it trades information against distance over a cost valley while respecting obstacles and a time budget.
- 2papers (IFAC 2022, Knowledge-Based Systems 2025)
Why greedy is not enough
A planner that only picks the best next sample walks into dead ends: a narrow inlet, an island, a front that bends behind the vehicle. Over a few hours that short-sightedness wastes most of the battery.
The approach
- A cost valley. Information value, distance, obstacles and the remaining time budget are folded into one cost field. The good routes run along the bottom of the valley.
- RRT* over it. A random tree grows through free water and rewires itself towards cheaper routes, so the vehicle commits to a path that is good for the whole horizon, not just the next leg.
- Replanning. As new measurements change the belief, the valley changes, and the plan follows.
The first version was published at IFAC CAMS 2022, and the full cost-valley method in Knowledge-Based Systems in 2025.