Project
UAV Routing
Maximum Information Collection under Time Windows and Energy Consumption
Overview
A fixed-wing UAV visits surveillance targets whose information value changes with the time of arrival, under hard time windows and a nonlinear propulsion-energy budget. The model chooses which targets to visit, in what order, the flight speed on every leg, and how long to loiter. The resulting mixed-integer nonlinear program is reformulated exactly as a mixed-integer second-order cone program through a chain of three rotated cones, and solved at scale by an iterated local search whose evaluation oracle is a small SOCP over a fixed route.
Search Replay
An interactive replay of the matheuristic on the benchmark instances. It steps through every route the search accepts, showing the flown tour, the targets it has not scheduled, the candidate moves it weighed at each step with their selection probabilities, and the shake steps that pull it out of a local optimum. Pick an instance from the menu to replay it.
Open Search ReplayCode
The model, the matheuristic and the experiments are open source.
View on GitHub