Perspective on SDSB Human Visual Knowledge and Intelligence for Happiness Campus

Suffering globally by COVID-19 since 2020 constrained learner and worker outdoors, of which campus and public area naturally met environmental protection issue. In such cases, a newly AI moveable application, naming Self-Driving Sweeper Bot (SDSB), is invented by intelligently coordinating between self-driving system and sweeper mechanism. In this paper, the perspective on SDSB in terms of human visual knowledge and intelligence between pedestrian security and sweeping efficiency in campus is reported. To reach such a goal, our investigation is shown that human visual knowledge and intelligence, played a critical role requiring routinely collecting and learning visual dataset, accompanied with optimizing procedure by exploring the object recognition methods e.g., CNN, R-CNN, Fast-RCNN and Yolo, for detecting campus objects (including, pedestrians, vehicles, common rubbishes, i.e. fallen leaves, waste papers, plastic bottles etc.), and image segmentation techniques e.g., U-net for constraining sweeping road. In the preliminarily experiments, observation is shown that the factors for object detection and road segmentation in terms of weather, sunshine direction and shadowing/non-shadowing by trees and facilities are highly influencing on SDSB visual intelligence.