面向复杂环境公共安全防控的移动机器人自主作业过程快速3D-SLAM技术

肩负复杂环境公共安全防控的智能移动机器人需要面对复杂多变、甚至可能存在危险的场景进行稳健高效的自主作业，而高性能的三维同步定位与地图构建（3D-SLAM）则是必要的技术保障。因此，面向复杂环境的3D-SLAM技术已成为智能机器人领域的前沿竞争性研究热点。本项目针对复杂环境下移动机器人在公共安全防控过程中自主作业的工程背景和技术需求，将FastSLAM的开放式组织架构同箱粒子滤波、机器学习、群智能优化、以及随机有限集等技术工具有机地结合起来，主要研究：面向非结构化复杂环境的非参Bayes学习进化式Gauss箱粒子滤波快速3D-SLAM；面向强干扰复杂环境的势均衡多目标多Bernoulli指示器式箱粒子滤波快速3D-SLAM；并实现高性能快速3D-SLAM算法体系的综合集成与优化，进而研制复杂环境公共安全防控的移动机器人自主作业快速3D-SLAM应用软件，结合典型案例完成实体测试和性能评价。

The intelligent mobile robot which shoulders public security prevention and control in complex environment, needs to face the complex and changeable scenes, or even potentially dangerous scenes for robust and efficient autonomous operation, while the high-performance 3D simultaneous localization and mapping (3D-SLAM) is the necessary technical support. Therefore, technology of 3D-SLAM for complex environment has become a frontier competitive research hotspot in the field of intelligent robot. Aiming at the engineering background and technical requirements of autonomous operation of mobile robots in the process of public security prevention and control in complex environment, this project will organically combine the open organizational structure of FastSLAM with the technical tools such as box particle filter, machine learning, swarm intelligence optimization, and random finite set, the major researches are organized as follows: fast 3D-SLAM algorithm based on the non-parametric Bayesian learning evolutionary for Gauss box particle filter towards unstructured complex environment; fast 3D-SLAM algorithm based on cardinality balanced multi-target multi-Bernoulli filter with indicator box particle towards strong disturbed complex environment; and the implementation of synthetic integration and optimization of high-performance fast 3D-SLAM algorithm system, to further develop fast 3D-SLAM application software of autonomous operation of mobile robot for public security prevention and control in complex environment, combining typical cases to complete the entity test and performance evaluation.