This paper introduces SLATO as a set of artificial intelligence traffic congestion mitigation and timing optimization methods that support manual training. The ideas of this paper include built-in traffic flow and micro-simulation models, real-time reception of traffic status data, and the acceptance of manually set parameters for network structure characteristics and traffic control demands. It uses artificial intelligence methods to conduct real-time traffic status analysis and decides whether to extend the phase time, adopting a second-by-second signal timing method. The paper first elaborates on the technical framework of SLATO, detailing the four main intelligent processing systems within the framework: state perception, dual-loop optimization, effect evaluation, and phase operation, and then explains the operating mechanism and principles of SLATO. As a key technical difficulty, the paper discusses urban-level traffic congestion control strategies separately, combining SLATO's distributed node optimization and urban-level central coordination dual-loop optimization logic. Finally, the paper briefly discusses the current research and application progress of SLATO, as well as its development prospects.