The invention provides a vision-based multi-industrial-
robot fault detection method and
system. The vision-based multi-industrial-
robot fault detection method comprises the following steps of S1, collecting a multi-industrial-
robot standard operation video, establishing multiple single-industrial-robot standard
operation mode video frame sequences A13, and executing S2; S2, acquiring a multi-industrial-robot operation video in real time, establishing multiple single-industrial-robot real-time operation video frame sequences A22, and executing S3; S3, matching a single-industrial-robot real-time action image with an image in a corresponding single-industrial-robot standard
operation mode video
frame sequence A13, detecting whether the single
industrial robot acts abnormally or not by adopting a two-stage method, if so, executing S4, and if not, executing S2; and S4, controlling the
industrial robot to stop suddenly. The method and the
system have the advantages that sudden faults of anindustrial robot body are found in a non-
contact mode, so that the safety accident that the robot hurts people during man-
machine cooperation is avoided, and the detection process is simple and accurate.