Abstract: A micro-robot magnetic drive device and a control method based on double closed loop three-dimensional path tracking are disclosed. The method includes: inputting a desired tracking path, obtaining current pose information of a magnetic micro-robot through a camera, and then calculating a position of a center of mass, an actual axial direction, coordinates of a desired position point with the shortest distance from the center of mass on a desired tracking path, and a tangent direction of this point; calculating a horizontal distance, a vertical distance, a direction angle error, and a pitch angle error of the two points according to the actual axial direction, the tangent direction, and disturbance compensation; and obtaining a required rotating magnetic field according to a designed position closed loop controller.

Abstract: A method and apparatus for training a compound property prediction model, a device, a storage medium and a program product. A implementation of the method comprises: acquiring an unannotated compound data set; pre-training a graph neural network using the unannotated compound data set to obtain a pre-trained graph neural network; acquiring a plurality of annotated compound data sets, each annotated compound data set being annotated with one kind of compound property; and performing multi-task training on the pre-trained graph neural network using the plurality of annotated compound data sets, to obtain a compound property prediction model, the compound property prediction model being used to predict a plurality kinds of properties of a compound.