随着人工智能、电子、传感器和控制等技术领域的发展，多智能体系统已具备自主协作完成特定任务的潜力。而不依赖于外部基础设施的相对/绝对定位与时间同步技术是支撑多智能体系统在未知环境下协作完成任务的基础。进而，当多智能体系统具有群体密集和自主移动的特征时，其对定位与时间同步技术的精度、频度、实时性和支持容量提出了更为严格的要求。本文针对密集移动多智能体系统对自主确定时空信息的迫切需求，基于无线测距定位技术和多传感器融合技术，开展多智能体协同定位与时间同步的关键技术研究，以期实现密集移动多智能体系统绝对/相对时空基准自主建立和保持的核心能力突破。研究成果具体包括：首先，提出了一种面向密集移动多智能体系统的基于主从分层架构和无线广播式定位体制的协同定位与时间同步方案BLAS。该方案在统一的无线定位协议框架下，既可支持有限数量主智能体的自主定位与时间同步以建立和保持相对时空基准，又可为不限数量的从智能体提供该时空基准下的无源定位与时间同步，有效降低了密集移动多智能体定位与同步对无线空口资源和节点计算能力的要求。第二，在BLAS方案下，提出一种融合主智能体间测距值和内部传感器的协同绝对定位方案从而实现多智能体绝对空间基准的建立。通过分析系统可观测性，推导出绝对空间基准完全可观测条件。进而设计了融合超宽带测距和轮式编码器的协同绝对定位系统原型，验证了所提方案的有效性和理论分析的正确性。第三，在BLAS方案下，结合对未知环境进行感知的需求，提出一种融合主智能体间测距值和视觉传感器采集图像的协同同时定位与建图方案，从而在实现多智能体绝对空间基准建立的同时也可高效地完成未知环境的三维建图。进而设计了融合超宽带测距和单目相机的原型系统，验证了所提方案能够实时准确估计出所有主智能体在环境中的绝对位置并创建真实尺寸的环境地图。第四，在BLAS方案下，提出了一种从智能体分布式三步联合定位与时间同步算法，以提升从智能体在主智能体实时建立的存在误差的时空基准下的时空参数估计精度。由于充分考虑了主智能体自建时空基准的不确定性，所提算法定位与时间同步精度显著优于现有方法，在小噪声条件下可达到理论界性能，并在从智能体钟差较大时具有更好的鲁棒性。最后，设计实现了密集移动多智能体系统外场实验软硬件平台，通过实验验证了所提BLAS方案的可行性以及相关状态估计算法的有效性。

Along with the development of artificial intelligence, electronic, sensor, and control technologies, the multi-agent system (MAS) has the potential to autonomously and collaboratively perform certain tasks in unknown environments. Relative/absolute localization and time synchronization that do not require external infrastructures are the supporting technologies for the MAS to complete tasks in unknown environments. Furthermore, when the MAS has the characteristics of density and mobility, more stringent requirements are proposed to the accuracy, frequency, real-time performance and system capacity of the localization and time synchronization technologies. In this paper, facing the urgent demand for the autonomous determination of spatiotemporal information in the dense moving MAS, based on wireless ranging and positioning and multi-sensor fusion technologies, the key technology research of multi-agent collaborative localization and time synchronization is carried out, to break through the dense moving MAS's core capacity of autonomous absolute/relative spatiotemporal reference establishment and maintenance. The main contributions are summarized as follows. Firstly, a collaborative localization and synchronization scheme for dense moving MAS based on leader-follower layering and wireless broadcast architecture, namely broadcast localization and synchronization (BLAS), is proposed. Under a unified wireless positioning framework, BLAS cannot only support the autonomous localization and time synchronization for a limited number of leader agents to establish and maintain a relative spatiotemporal reference, but also support passive localization and time synchronization for an unlimited number of follower agents with respect to the spatiotemporal reference. In this way, the radio resource and node computing power for localization and time synchronization in the MAS can be effectively reduced. Secondly, under the BLAS scheme, a collaborative absolute localization scheme based on ranging and the internal sensor is proposed to establish absolute spatial reference in the MAS. The observability properties of the system are investigated and the condition of full observability of the system is derived. Furthermore, a collaborative absolute localization prototype system based on ultra-wideband (UWB) and wheel encoders is proposed. The feasibility of the proposed scheme and the correctness of theoretical analysis are verified. Thirdly, under the BLAS scheme, consider the need of unknown environment perception, a collaborative simultaneous localization and mapping scheme utilizing ranging measurements and visual images is proposed, so that the MAS can simultaneously establish absolute spatial reference and effectively build the map of unknown environments. Furthermore, a prototype system combining UWB and the monocular camera is designed, which verifies that the proposed scheme can estimate the absolute position of all leader agents and build a real-size environment map accurately and in real-time. Fourthly, under the BLAS scheme, a distributed three-step joint localization and time synchronization algorithm is proposed to improve the accuracy of the estimation of spatiotemporal parameters in the presence of spatiotemporal reference uncertainties. By fully considering the uncertainties in spatiotemporal reference self-established by leader agents, the proposed algorithm performs better than existing methods in terms of localization and synchronization accuracy. Besides, it can reach the Cram\'er-Rao lower bound (CRLB) under small noises and is robust to the follower agent with a large clock offset. Finally, the software and hardware evaluation platform of dense moving MAS is designed and implemented, the experimental results verify the feasibility of the proposed BLAS scheme and the effectiveness of the related state estimation algorithms.