Work

Intelligent traffic management with Mamba-ATSP: Optimizing signal control and path planning using multi-modal data fusion

Year: 2026

Type: article

Abstract: Efficient processing of multi-source data and dynamic environments is essential for addressing the complex multi-task challenges in intelligent traffic signal control and path planning. Traditional models cannot capture time dependence, global information, and non-linearity relationships inherent in traffic flows. Optimization performance is limited. for dealing with these challenges, we introduce Mamba-ATSP, self-attention mechanism and manaba framework are integrated. Mamba-ATSP In spatial time management, traffic signal optimization and route planning are characterized by modules that integrate multi modal data mission planning and ego attention. Experimental results demonstrate that Mamba-ATSP outperforms established models, such as DeepTraffic, ST-Mamba, and LightPath, on the PEMS and HighD datasets, achieving superior accuracy and efficiency across key metrics like traffic flow, path selection time, average waiting time, and real-time responsiveness. Ablation studies underscore the significance of each module, emphasizing the crucial roles of multi-modal fusion, self-attention, and the Mamba framework. Mamba-ATSP not only does it improve traffic signal control and operation planning, but also shows strong adaptability in dynamic traffic environments and extends its potential to other intelligent traffic systems.

Source: Alexandria Engineering Journal

DOI: 10.1016/j.aej.2026.01.010

Authors: Xue Li, Zheng Li, Yan Wang

Institutions: Shandong Jiaotong University, Central China Normal University

Language: English

Topic: Traffic control and management

Subfield: Control and Systems Engineering

Field: Engineering

Domain: Physical Sciences

SDGs: Industry, innovation and infrastructure, Sustainable cities and communities

Open Access status: gold

Estimated APC paid: $1,900