Abstract: To address problems of overshoot during path tracking control and unstable driving of agricultural machinery caused by polygonal trajectory curvature mutation in hilly and mountainous areas, an arc switching method for continuous curvature change was proposed．Based on six path distributions, optimal path switching positions were determined using broken lines' intersection points and the minimum turning radius of agricultural machinery．Mathematical models were established to determine circular arc trajectories．On this basis, a model predictive control（MPC）system model based on Matlab/Simulink-Carsim was built．Verification was carried out using different polygonal angles（60°, 90°, 120° and 150°）and different forward speeds（0.5 and 1.0 m/s）．Verification results showed that when agricultural machinery moved forward at speeds of 0.5 and 1.0 m/s, compared with traditional method before optimization, optimized method reduced average lateral deviation by 36.79% and 39.46%, respectively, and reduced average heading angle deviation by 35.91% and 30.43%, respectively．In addition, both the maximum lateral deviation and heading angle deviation reduced．Path tracking accuracy and driving stability of agricultural machinery were significantly improved, better meeting needs for polygonal paths tracking．