The majority of “hot car deaths” occurs when a child is left unattended in their car, and the death toll is increasing as the number of car owners are increasing. There is a growing need for a technology that can prevent children from dying in hot cars by effectively detecting their presence and alerting the parents and/or by-standers if any child is left unattended inside a car. We have developed an In-Vehicle Life Presence Detection System to identify the presence of children in vehicles and provide alerts to vehicle users or third-party services. The life presence detection system consists radar detection module that utilizes a 60Ghz mmWave radar to detect in-vehicle object and the presence of life. The system’s wireless transmission module, leveraging on NB-IoT technology, relays detected passenger(s) information and status from the radar and alert vehicle owners or responsible authorities on their mobile phone using a mobile application. The detection solution combines both dynamic and static motion detection. Dynamic motion detection is based on large body movements and static motion detection is based on vital signal detection such as movements caused by breathing and heartbeat. In the process of static motion detection, the extraction of vital signals and spectrum estimation information are the focus of our research. According to the 200 tests conducted in a sedan car, our results demonstrate a 100% detection accuracy rate for single or multiple passengers. Furthermore, the system did not produce any false alarms when no passengers were present in the vehicle. Based on our research, we have identified that there is room for improvement in the detection accuracy of in-vehicle life classification. This task poses a challenge due to the overlap in respiratory and heartbeat rates between humans and pets such as cats and dogs. To improve accuracy, additional parameters are needed to distinguish between humans and pets. One such parameter could be the use of a millimeter wave radar with higher resolution to accurately detect the size of the life form. We developed a method to eliminate errors caused by random body movement during vital sign detection. The advantages of this method are as follows: (1) it can simultaneously detect the life characteristics of multiple targets within the radar's field of view; (2) it corrects the detection errors caused by the random body movements of the target during the radar detection process, which is more suitable for complex and changing real-world scenarios than cooperative radar detection methods; (3) the sequential variational mode decomposition algorithm is used to extract the respiratory and heartbeat signals of the target, which does not require setting the number of modes in advance, and has a faster convergence speed compared to the variational mode decomposition algorithm. Overall, our In-Vehicle Life Presence Detection System has provided the reliable life detection function to avoid heatstroke-related injury and death by the radar sensor and detection technology with enough sensitivity to detect dynamic motion and vital sign, and the reliable NB-IoT technology and mobile application for alarm transmission.

Mr. CHULIANG SHAN, Consultant, Hong Kong Productivity Council