Work

A controlled evaluation of objective-responsive genetic-operator adaptation for variable-length 3D UAV path planning

Year: 2026
Type: article
Abstract: Three-dimensional UAV path planning requires simultaneous satisfaction of obstacle-clearance, altitude, and turning constraints. This study evaluates objective-responsive genetic-operator control within a variable-length waypoint representation. A common signal derived from the best penalized objective and a geometric lower bound adjusts selection, crossover, mutation, and elitism separately or jointly. Eight GA variants were compared in 960 runs across four synthetic environments using 30 matched seeds per environment and 100,000 objective evaluations per run. Fixed-length GA, particle swarm optimization, differential evolution, and a Dubai urban case extended the evaluation. Adaptation of individual operator families achieved the highest observed feasibility in three synthetic environments, although its pairwise feasibility advantages over the balanced GA did not survive correction for multiple comparisons. Full adaptation produced significantly worse final objectives than the balanced GA and three variants adapting individual operator families in Environment 3. In the two larger environments, median feedback signals remained near 0.21–0.22 as infeasibility penalties dominated, limiting changes in operator settings. Differential evolution with the minimum fixed waypoint count achieved full feasibility in all synthetic environments; selection, mutation, and elitism adaptation each yielded 10/10 feasible outcomes in Dubai. These results connect adaptation outcomes with the measured feedback response and demonstrate the importance of evaluating operator control jointly with feasibility and trajectory dimension.
Language: English
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7233549601
FWCI:
Cites:
Cited by:
Related to: