Even though state of the art technologies in assisting drivers to prevent accident or to minimize damages are currently adopted in most vehicles, few methodologies have been proposed in terms of evaluating the effectiveness and performance of those systems being operated in local roadway environment.

Those assistance systems would warn drivers' lack of awareness or carelessness through proper warning signal but fault or misjudgment in detecting an incidence often discourage customers and drive them away from relying on relevant systems.

The blind spot detection (BSD) system has been designed to warn drivers when adjacent vehicles are out of driver visible sight with close proximity.

At a certain curvature lane, even though a vehicle follows preceding one in the same lane, it would be located in the warning zone and might provoke an alarm.

A study objective is to evaluate the characteristics of BSD system in various curvature roads.

The roadway design standard with flat surface eliminating the effect from elevation is used in introducing curvature road. The vision based BSD system with lane width of 3.5m, lane radius of curvature as 110 and 125m is used. The BSD system detection criteria are applied referring to ISO 17387. The possibilities of system's misjudgment are based on the proportion of current lane area detected by the system within the detection range. The assessment of the area is found from analytical analysis or Monte Carlo method based on the geometry.

The BSD system performance is found to be characterized by lateral and longitudinal detection range, roadway width and curvature. The results show that the possibility of misjudgment increases as the curvature decreases. As the longitudinal detection range increases, erroneous measurement increases gradually and reaches sudden leap. The misjudgment trend with respect to lateral detection range shows similar variation.

The other parameters affecting the system performance might be current and following vehicle orientation, location, lane width, vehicle speed with different inter vehicle gap. Further study is encouraged.

Various calculation methods would be proposed for the assessment of overlapped area composed of lines and arc. Since the combination of geometrical disposition and operating condition would require detailed classification of cases for assessment purposes, more thorough studies are required. Different roadway design like clothoid curve would require complex evaluation method other than simple analytical approach.

A vision base BSD system might cause malfunction and deteriorate the credibility that the driver assistance system has been accumulating. As for the misjudgment of system, similar analogy can be applied to different type of BSD system like radar based one. Studied results can be used to enhance the system reliability by adjusting the lateral and longitudinal detection range. Proper test procedures considering roadway design criteria and vehicle condition can be proposed for evaluating BSD system performance.