A machine learning method includes: a first learning step which is performed in a phase before a neural network is installed in a mobile robot and in which a stationary first obstacle is placed in a set space and the first obstacle is placed at different positions using simulation so that the neural network repeatedly learns a path from a starting point to the destination which avoids the first obstacle; and a second learning step which is performed in a phase after the neural network is installed in the mobile robot and in which, when the mobile robot recognizes a second obstacle that operates around the mobile robot in a space where the mobile robot moves, the neural network repeatedly learns a path to the destination which avoids the second obstacle every time the mobile robot recognizes the second obstacle.

A system and method for classifying products manufactured via a manufacturing process. A processor receives an input dataset, and extracts features of the input dataset at two or more levels of abstraction. The processor combines the extracted features and provides the combined extracted features to a classifier. The classifier is trained based on the combined extracted features for learning a pattern of not-faulty products. The trained classifier is configured to receive data for a product to be classified, to output a prediction for the product based on the received data.

The invention relates to a method for operating a robot and to a robot, wherein the robot comprises movable elements ELE.sub.m which can be driven by actuators AKT.sub.n, and is designed to carry out a movement B with the elements ELE.sub.m, and wherein the robot comprises a detection system for determining signals W.sub.G.sub.k.sub.B(t) of a group of measurement variables G.sub.k.sup.B characterizing the movement B of the elements ELE.sub.m and the interactions thereof with an environment. The proposed method comprises the following steps: determining (10), by means of the detection system, reference signals W.sub.G.sub.k.sub.B.sup.R(t) of the measurement variables G.sub.k.sup.B during at least one execution of the movement B of the elements ELE.sub.m which is in the form of a reference movement B; automatically determining (102), based on the reference signals W.sub.G.sub.k.sub.B.sup.R (t), using an adaptive method, a mathematical model M.sub.G.sub.k.sub.B for describing the reference movement B including the reference interactions by the measurement variables G.sub.k.sup.B, during a normal execution of the movement B by the model M.sub.G.sub.k.sub.B; predicting (103) signals W.sub.G.sub.k.sub.B.sup.P(t) for describing the reference movement B, including the reference interactions by the measurement variables G.sub.k.sup.B; comparing (104) the signals W.sub.G.sub.k.sub.B(t) determined currently during the normal execution of the movement B with the predicted signals W.sub.G.sub.k.sub.B(t) for determining a deviation Δ.sub.G.sub.k.sub.B(t) between W.sub.G.sub.k.sub.B.sup.P(t) and in W.sub.G.sub.k.sub.B; insofar as the deviation Δ.sub.G.sub.k.sub.B(t) does not meet a predefined condition BED.sub.G.sub.k.sub.B, based on the deviation Δ.sub.G.sub.k.sub.B(t) classifying (105) the current deviation Δ.sub.G.sub.k.sub.B(t) in one of a number I of predefined error categories F.sub.i,G.sub.k.sub.B(Δ.sub.G.sub.k.sub.B(t)), wherein predefined control information S.sub.F.sub.i.sub.,G.sub.k.sub.B(t) for the actuators AKT.sub.k is produced for each of the error categories F.sub.i,G.sub.k.sub.B(Δ.sub.G.sub.k.sub.B

A system and method for classifying products manufactured via a manufacturing process. A processor receives an input dataset, and extracts features of the input dataset at two or more levels of abstraction. The processor combines the extracted features and provides the combined extracted features to a classifier. The classifier is trained based on the combined extracted features for learning a pattern of not-faulty products. The trained classifier is configured to receive data for a product to be classified, to output a prediction for the product based on the received data.

A machine learning method includes: a first learning step which is performed in a phase before a neural network is installed in a mobile robot and in which a stationary first obstacle is placed in a set space and the first obstacle is placed at different positions using simulation so that the neural network repeatedly learns a path from a starting point to the destination which avoids the first obstacle; and a second learning step which is performed in a phase after the neural network is installed in the mobile robot and in which, when the mobile robot recognizes a second obstacle that operates around the mobile robot in a space where the mobile robot moves, the neural network repeatedly learns a path to the destination which avoids the second obstacle every time the mobile robot recognizes the second obstacle.