PAANI: On-Device AI for Low-Cost River Monitoring Robots

PAANI is an on-device perception-to-guidance architecture for river monitoring robots, combining a YOLO11n detector and a MobileNetV3 Small semantic segmenter with timestamp-aligned evidence fusion, running on Arduino UNO Q with ROS 2 interfaces.

Instead of relying solely on geographic waypoints, PAANI visually interprets obstacles and water boundaries, and provides explainable guidance that exposes contributing evidence and policy reasons, while flagging unreliable evidence.

The detector achieved mAP@0.5 of 0.7388 and the segmenter mIoU of 0.9750, with median latency of 467.8 ms at 0.5 Hz and a model size of 14.817 MB.

Yes, the architecture is reusable as an edge robotics foundation, but note the study distinguishes model accuracy and on-board execution from validated on-water collision avoidance, so additional field testing is required.