Mission-Driven UAV Path Selection: Post Hoc Cost Evaluation of Deterministic and Sampling Approaches

Unmanned Aerial Vehicles (UAVs) are increasingly deployed in hazardous and dynamic environments, where path planning requires balancing competing objectives beyond simple distance minimisation. Classical planners such as Dijkstra, A*, and RRT* generate paths efficiently but often overlook mission-specific trade-offs involving energy use, risk avoidance, and reward maximisation. This work proposes a unified evaluation framework that integrates grid-based (Dijkstra, A*, weighted A*) and sampling-based (RRT, CARRT*) planners within parameterised environments embedding a range of functions into penalty and reward zones. A global cost function, J = αL + βE + γp− δR, is applied post hoc to decouple path generation from mission prioritisation, enabling rapid reassessment under changing objectives such as low-fuel, high-safety, or speed-priority scenarios. Experiments conducted on an Apple M2 CPU, repeated three times per configuration to ensure statistical robustness, demonstrate that CARRT* achieves the lowest mission costs and highest efficiency for fuel- and time-sensitive missions, while deterministic grid-based planners perform better in safety- and reward-oriented contexts in four environments. These findings indicate that optimal UAV path planning depends not only on algorithmic efficiency but also on aligning planner choice with mission priorities. The framework provides a reproducible methodology for benchmarking and deploying mission-aware path planning strategies in research and operational settings.