The invention belongs to the technical field of robot positioning, and mainly relates to an SLAM loopback detection method based on a mobile robot. The SLAM loopback detection method comprises the following steps of: S1, carrying out laser preprocessing on laser motion distortion generated by the motion of the robot, and extracting the edge point and plane point features in a distance image; S2, converting relative attitudes of continuous frames to obtain a radar odometer, and forming a local sub-graph through selection of key frames; and S3, adopting a brand-new descriptor EPSC to represent local sub-graph features to carry out loop-back detection. According to the SLAM loopback detection method, a brand-new descriptor EPSC is adopted for loop detection, drift errors can be effectively reduced, a global map with more optimized consistency is established, and the closed-loop detection precision, recall rate and calculation efficiency are improved.