As collaborative robots share workspace with humans, physical collision between the human and the robot is inevitable. In order to ensure the safety of humans an international standard, ISO/TS 15066, presents the collision peak pressure which should not be exceeded in the case of collision. To prevent human injury from collision, robots should be aware of its potential risk by predicting the collision peak pressure in real time. The pressure varies depending on the shape of the impactor and the contact area in the event of a collision, the collision peak pressure needs to be analyzed by considering the contact surface nonlinearity. Furthermore, the deformation of human skin exhibits nonlinear elastic behavior. Modeling the human-robot contact behavior with the finite element method would be a conventional approach for solving complex nonlinearity. However, it is not practically possible to perform real-time evaluations because of its high computational cost. This paper introduces a real-time pressure estimation method using a mathematical model for evaluating collision safety for a collaborative robot. To verify the effectiveness of the real-time model associated with the contact surface, impactor shape and the material, we compare with a FE model based calculation results in this paper. The relationship between accuracy and the computational cost is also described.