Vehicle-road collaboration is inseparable for smart cities, and the potholes detection is an essential direction in vehicleroad collaboration. With the development of artificial surface observation technology and surveying technology, road potholes detection accuracy has improved in recent years. However, the legacy detection methods do not have the ease of service and realtime observation capability, and thus the potholes in the road cannot be mapped in time. To solve this key issue, we proposed a reflectometry method to realize real-time potholes observation with vibration signals analysis and Spatio-temporal trajectory fusion. We further developed several prototype devices for validation. These prototype devices measure the acceleration signal mounted on the wheel steering lever and implement edge signal processing and Spatio-temporal information fusion on the prototype. Observation results and Spatio-temporal information are been rapidly transmitted to the sensing server via NB-IoT. Results and analyses demonstrated that this method successfully enabled the potholes observation in real-time through light and a rapidly deployable platform that relies on repeated trajectory data from the vehicle. Results also demonstrated that this method was not limited by vehicle type, speed, or engine operating condition. In the road experiment, the proposed method provided stable, efficient, and real-time potholes observation results. Compared with traditional methods, the method reduces costs and improves sensing efficiency based on Spatio-temporal trajectory fusion, and potholes information can be prompted in real-time. This innovation provides a strategic exploration and thinking to deal with road potholes in real-time sensing.