The high-precision positioning in the whole workspace domain is the key to realizing the full-range precision operation of large workspace robots. To improve the reliability and spatial adaptability of parameter calibration in the full workspace domain of the robot, this article studies the non-probabilistic quantification methods of many uncertainties in the robot itself, analyzes the difference of the influence of uncertain parameters on the end positioning accuracy of the robot in different workspace domains, and partitions the whole workspace domain according to the non-probabilistic reliability index of positioning accuracy. A non-probabilistic reliability calibration method for robots under the framework of workspace partition is proposed. The example shows that, after partition calibration compensation, the average lower and upper bounds of error intervals in x, y, and z directions decrease by 40. 16% , 59. 36% , and 59. 08% , 40. 87% , as well as 54. 24% , 33. 98% , respectively. Moreover, the compensated robot has a fast response speed and small fluctuation during movement. It is proved that the proposed method is effective in reducing the end error range in the whole working domain, improving the absolute positioning accuracy of the robot and the spatial adaptability of the calibration.