Sensor Path Planning and Scheduling for Aircraft Emergency Landing Field Monitoring

This paper presents a sensor path planning and scheduling approach for the monitoring of a series of aircraft emergency landing sites with a single on-board sensor. Since the detection of on-site hazards and obstacles requires a high ground sample distance and therefore a low field of view of the electro-optical camera, it is unlikely to capture multiple sites in one view. A scheduling algorithm producing an observation sequence to maximize hazard detection performance and to minimize the time of switching between multiple landing sites while respecting priority requirements is required. We propose an algorithm creating such sequences by combining the generation of a sensor path connecting all viable landing sites with a sensor scheduler coordinating the gimbal control and the sensor path execution. The quality estimation of the sensor path is based on three different factors: the predicted sensor perception performance, the spatial distance between the different landing sites and the landing site priority rating. The sensor path execution of the sensor scheduler depends on changes of the sensor perception performance prediction, due to positional changes of the gimbal, as well as of the landing site priority rating. To find a suitable algorithm for the sensor path planning, we compare different heuristics, a metaheuristic as well as an optimization-based approach regarding the quality of the resulting sensor path and the time needed for its generation. This evaluation is performed on a set of emergency landing site configurations generated from simulated flights on specific flight levels. The results show the most suitable algorithms with respect to both factors, allowing for the selection of implementations with the best compromise for our use case.