Core Concepts
The author proposes a sampling-based approach to address risk-aware path planning around dynamic engagement zones, highlighting the limitations of existing deterministic methods and optimal-control strategies.
Abstract
The study introduces a novel sampling-based method for risk-aware path planning in the presence of dynamic engagement zones. Existing methods relying on calculus of variations are criticized for scalability issues and sensitivity to initial guesses. The proposed algorithm leverages Rapidly-exploring Random Trees (RRT*) to navigate through a large number of engagement zones efficiently. By transforming dynamic two-dimensional obstacles into three-dimensional static obstacles, the algorithm aims to find feasible flight plans for vehicles in threat-laden environments. The Monte Carlo experiment evaluates the success rate and average path length based on varying numbers of engagement zones and computation time available to the planner.
Stats
"A total of N = 16 EZ regions were randomly generated in the operating domain."
"For each number of EZs, N ∈ {4, 8, 20, 24}, a total of M = 500 different scenarios were generated."
"The success rate varied from 96.8 % to 99.6 % with τsolve = 120 s."
"The success rate for the solver with τsolve = 160 s is 100%."
"The success rate for the solver with τsolve = 5 s ranges from 85.4% to 99.0%."