This paper introduces the use of Visible Light Communication (VLC) to enhance traffic signal efficiency and vehicle trajectory management at urban intersections. By integrating VLC localization with learning-based traffic signal control, a multi-intersection traffic system is proposed. VLC enables communication between connected vehicles and infrastructure through mobile optical receivers. The primary objectives are to reduce waiting times and improve overall traffic safety by accommodating diverse traffic movements during multiple signal phases. Cooperative mechanisms and queue/response interactions balance traffic flow between intersections, enhancing road network performance. A reinforcement learning scheme optimally schedules traffic signals, with agents at each intersection using VLC-enabled vehicle communication to improve traffic flow and overall system optimization. Evaluated using the SUMO urban mobility simulator, the system demonstrates reduced waiting and travel times. The decentralized and scalable nature of this approach highlights its potential applicability in real-world traffic scenarios.