Abstract: An intelligent prediction algorithm to be introduced in conventional hybrid position/force control model is developed to deal with the force-controlled execution of compliant robot tasks in an unknown environment. The algorithm makes use of the three factors to predict and readjust the expected trajectory given in hybrid position/force control model in future sampling, with the environmental change in curvature and stiffness taken into account. To verify the effectiveness of the proposed algorithm, an open architecture platform of robot force control is set up to investigate experimentally the force control on irregularly curved surfaces for different expected forces and different tracking velocities. The influences of unknown environmental parameters on the contact force are analyzed on the basis of experimental data. The experimental results verified that the way proposed for intelligent prediction has high adaptability to unknown environment in changing. The error of force controlling is within 3% in constrained motion.