This study investigates the integration of augmented reality (AR) into engineering education, with a specific focus on its impact on spatial visualization abilities and the cognitive load in the context of an engineering drawing course. The research employs a structured approach, involving three experimental groups utilizing marker-based (MBAR), markerless (MLAR), and web-based AR (WBAR) applications, in comparison to a control group receiving traditional instruction. The primary objective is to rigorously assess the comparative effectiveness of these instructional approaches in enhancing spatial visualization skills and reducing the cognitive load of undergraduate engineering students. Spatial visualization, a vital cognitive skill for engineering disciplines, is evaluated using the Purdue Spatial Visualization Test: Rotations (PSVT-R) questionnaire. The cognitive load of students is assessed with the help of a questionnaire comprising items related to mental load and effort. This study not only quantifies the impact of AR applications but also highlights subtle differences among the various AR modalities. The findings suggested that the group that used marker-based AR had better improvement in their spatial ability and felt less cognitive load compared to those who utilized markerless AR and web-based AR.