Abstract: Aiming at the problem of insufficient shock absorption comfort of wheelchairs during the travel of the elderly and disabled, this paper proposed a lightweight and fast - response active suspension system （LARSS）. The system integrated LiDAR and binocular cameras to obtain environmental information, used 3D - CNN and vector embedding algorithms in the deep - learning technology system to process data, constructed a ground height cloud point map, analyzed road surface features with the fuzzy PID algorithm, and drived the active suspension to achieve millisecond - level dynamic response. Practice showed that this system effectively enhanced the shock absorption performance under complex road conditions, realized the efficient linkage of environmental perception, data processing, and suspension control, created stable and comfortable travel conditions for special groups, and had significant engineering application potential.