Three-dimensional path planning for source seeking UAV based on multistrategy dung beetle optimization algorithm

Abstract

In outdoor radioactive leakage scenarios, unmanned aerial vehicles (UAVs) can replace humans in performing radioactive source search tasks, but three-dimensional path planning faces challenges such as runtime inefficiency and premature convergence. The intelligent bionic dung beetle optimization (DBO) algorithm offers a plausible resolution, but it suffers from issues such as vulnerability to local optima stagnation, insufficient global search capability, and poor convergence precision. To ameliorate these shortcomings, this study devises a novel-enhanced multistrategy dung beetle optimization algorithm (MDBO). First, the Bernoulli–Tent chaotic initialization mechanism is adopted for the population to enhance solution variety. Second, a dynamic spiral search strategy balances local improvement and global discovery during breeding and foraging phases. Finally, adaptive Gaussian–Cauchy hybrid mutation perturbs optimal positions to expand the search scope. CEC2017 test results demonstrate that, compared with four other prominent intelligent optimization algorithms, the proposed algorithm achieved 11 optimal solutions and ranked first. Moreover, when compared with three outstanding improved DBO algorithms from the past 2 years, it also attained 19 optimal solutions and secured the first ranking. In the three-dimensional path planning experiments, MDBO achieved better fitness than the original algorithm across three obstacle scenarios. The average path cost was reduced by 2.52%, 2.30%, and 3.94%, respectively, while the optimal path cost decreased by 1.42%, 1.38%, and 1.94%. The standard deviation was reduced by 58.34%, 60.92%, and 39.90%, respectively. It consistently generated safe and optimal paths, verifying its effectiveness and practicality.