Research on multi-sensor fusion architecture for highway area hazard monitoring based on UAVs.

Highway area geohazards pose a serious threat to traffic safety, and traditional monitoring echniques struggle to meet the requirements for high precision and real-time capability, while existing unmanned aerial vehicles (UAVs) multi-sensor fusion schemes lack adaptability across diverse areas. Based on the monitoring requirements in different areas and variations in sensor performance, this study employs an improved AHP-TOPSIS method to conduct sensor fusion screening research. Firstly, an improved AHP method was adopted to convert qualitative requirements into quantitative weights; Secondly, integrating the improved TOPSIS method, based on the sensor performance scores after flying height correction and indicator weights, the single-sensor closeness was calculated, thereby screening and forming a single-sensor candidate set; Finally, the TOPSIS process was reapplied for all possible multi-sensor fusion schemes in the candidate set to evaluate these schemes and screen out the optimal fusion scheme for rural level, rural rolling, rural mountainous, and urban/suburban highway areas. Experimental results indicate: For rural level area highways, sensor fusion transitions from basic optical (RGB + IR) to refined optical combinations (RGB/Hyperspectral+Thermal infrared multispectral) as the height of UAVs increases; the rural rolling and mountainous area highways adopt RGB + LiDAR at low-to-medium heights (5-100m), while adjusting to Hyperspectral+SAR and LiDAR + SAR respectively at 100-120m to better adapt to monitoring requirements; urban/suburban area highways use RGB + IR at low-to-medium heights, while RGB+Thermal infrared multispectral proves more suitable at 100-120m heights. This study provides a highly adaptable, precise, and efficient technical scheme for highway hazard monitoring.