Abstract: Aiming at the problem of overcutting of cavity boundary reinforcing ribs, which is easy to occur in semi-closed compartment inner cavity robotic grinding and weight reduction, an online identification method of milling overcutting based on force feedback is proposed, which adopts a six-dimensional force sensor to monitor the change of milling force in real-time, and identifies the overcutting through the setting of feature thresholds. In order to solve the current problems of difficult to select the feature thresholds and low identification accuracy, a Kalman filtering algorithm is introduced for the real-time processing of the data to predict the trend of milling force change. Combined with the internal structural characteristics of the semi-enclosed section, a test platform is built for test verification. The results show that under the same milling parameters, the Kalman filter algorithm has a significant effect on the overcutting recognition accuracy, the overcutting amount is reduced by 70%–80%, and the overall control is below 0.2 mm, which meets the machining requirements and can effectively improve the yield of the workpiece. It provides a technical reference for solving the problem of weight reduction overcutting in semi-enclosed compartment inner cavity robot grinding.