The invention relates to a method and system for sensing grabbed object information through an artificial hand. The method comprises the steps that multi-dimensional grabbing pressure data generated when the artificial hand grabs an object and multi-dimensional shaking pressure data generated when the artificial hand shakes the object are obtained through a touch sensor; inputting the multi-dimensional grabbed pressure data into a DNN neural network to obtain one-dimensional quality feature data as quality information; s2, adding the one-dimensional quality feature data obtained in S2 to multi-dimensional shaking pressure data, and inputting the multi-dimensional shaking pressure data to an LSTM network to obtain friction information; and inputting the quality information and the friction information into a data label trained in advance to obtain the type of the grabbed object. Compared with the prior art, various information of the object is acquired by using a single touch source, and the method has the advantages of high sensing accuracy, low cost, high efficiency and the like.