摘要
针对现有屏蔽物环境下放射源搜寻算法匮乏的问题,本文提出了一种基于改进型同步扰动随机逼近(simultaneous perturbation stochastic approximation,SPSA)算法的组合驱动算法.该方法通过改进单边正交扰动向量,并引入动态步长调整策略,有效提升了算法的收敛性能;为了解决复杂环境下规划路径失效的问题,设计基于广度优先搜索的目标点修正机制,实现了障碍物环境下目标点的动态校正与可行性保证.针对屏蔽物存在下的多空间场景,提出探索状态切换机制,可主动探索未观测区域.为了评估所提算法在收敛性能、障碍物适应性和探索能力方面的性能,在 Gazebo 中建立了三个仿真环境,分别代表以下环境:开阔场地环境、复杂环境和辐射屏蔽环境.实验结果表明,该算法在收敛性能方面优于标准 SPSA 算法和粒子滤波算法,同时在复杂环境中也表现出良好的搜索能力.此外,当面临屏蔽环境时,本文所提算法展现出受阻后快速恢复对未知环境的主动探索能力.
In response to the challenges of radiation source localization in shielded environments,a combination-driven algorithm is proposed based on an improved simultaneous perturbation stochastic approximation(SPSA)framework.The proposed algorithm enhances the convergence performance by optimizing one-sided orthogonal perturbation vectors and incorporating a dynamic step-size adjustment strategy.To address path infeasibility in complex environments,a goal relocation mechanism based on breadth-first search(BFS)is introduced.For complex multi-space scenarios with shielding structures,an active exploration strategy is employed to proactively investigate unobserved regions.To evaluate the performance of the proposed algorithm in terms of convergence performance,obstacle adaptability,and exploration capacity,three simulation environments are set up in Gazebo,which respectively represent the following environments:open-field environment,complex environment and radiation shielding environment.The experiments demonstrate that the proposed algorithm outperforms the standard SPSA algorithm and particle filter algorithm in terms of convergence performance,while also exhibiting robust search capability in complex environments.Furthermore,when confronted with shielded environments,the proposed algorithm demonstrates the capability of actively exploring unknown areas within a relatively short period after being obstructed.