Alzheimer’s disease (AD) is characterized by progressive cognitive decline, with spatial memory impairment being one of its earliest symptoms. To support efficient and accurate behavioral assessments in rodent models of AD, we developed an automated, microcontroller-based radial arm maze integrated with eight infrared (IR) distance sensors. In contrast to traditional camera-based systems, our design eliminates the need for human intervention and visual tracking, offering consistent and reliable performance across diverse lighting conditions. The system enables the precise quantification of latency and short- and long-term memory errors. In validation experiments, rats in the AD group exhibited significantly longer latencies and higher memory error rates than sham controls beginning on Day 21 following Aβ1–42 injection, confirming the system’s sensitivity to cognitive decline. Additionally, spatial trajectory maps indicated more disorganized searching patterns and frequent pausing in AD rats. These results demonstrate the effectiveness of the proposed maze system as a robust, low-cost platform for automated cognitive assessment in preclinical AD research.