Dual-camera hue, saturation, value-based team robot detection for wheeled soccer

Abstract

The Indonesian Wheeled Soccer Robot Contest requires reliable team robot detection to support robot maneuverability and coordination during matches. This study presents a lightweight dual-camera vision system that combines hue, saturation, value (HSV) color segmentation, morphological processing, and contour filtering to identify cyan and magenta team markers. A 5 MP front-facing camera was used for frontal observation, while a 0.9 MP camera combined with an omnivision mirror provided panoramic coverage. The system was examined through operational detection observations under stable indoor illumination of approximately 250–300 lux and unstable illumination of approximately 150–250 lux. Under stable illumination, the front-facing camera detected the target at observed distances up to 8.0 m. Under unstable illumination, the maximum observed range decreased to 3.0 m, indicating the sensitivity of fixed HSV thresholds to illumination variation. The panoramic camera provided complementary short-range directional information but was affected by its lower resolution and mirror distortion. Because repeated-trial logs, frame-level annotations, and processing-time measurements were not retained, the study does not report statistical detection accuracy or quantified real-time performance. Adaptive threshold estimation and systematic computational benchmarking are recommended for future development.