The shared working space of multi-robot cooperative work enables the robots to interact with each other.This will inevitably lead to the collision of the robot during the movement process.Once the collision occurs,it will cause serious operational accidents.Thus,collision detection is one of the important parts of multi-robot collaborative technology.Aiming at multi-robot collision detection,collision detection based on discretization was mainly studied.The time step was divided on the path of planning,and then the static collision detection was continually called on the time step.Thus,the collision detection of the robot in the process of motion was completed.Aiming at collision detection between multi-robots based on ellipsoid modeling,ellipsoid was used to surround robot connecting rod,joint and other structures,through acquiring the minimum distance between ellipsoids to detect collision.To Solve the shortest distance between ellipsoids,the shortest distance algorithm based on ellipsoidal coordinate system was adopted,the local ellipsoidal coordinate system was established by taking the center of one ellipsoid as the origin and the length of the semi-axis of the ellipsoid,the shortest distance from another ellipsoid to the origin of the ellipsoidal coordinate system was solved by using the concept of isohedral surface,the computing efficiency was faster.In order to verify correctness of the collision detection algorithm,using the robotic tool box in MATLAB,the robot structure parameters were initialized,the spatial location parameters of the planned path were extracted,time steps were divided on the path,the collision detection algorithm was called repeatedly at each time step,to obtain the shortest distance of each discrete point in the process of robot motion.Finally,the shortest distance change curves were obtained,and the collision detection was carried out.The effectiveness of the proposed method was verified through a simulation example.