The study proposes an improved Monte Carlo algorithm based on random oscillation to calculate the workspace of a telescopic manipulator. In response to the uneven distribution of point cloud when using Monte Carlo algorithm to compute the workspace, we adopt an approach to divide a point cloud into distinct regions. Firstly, it segments the point cloud region by point cloud enveloping and spatial dilation. Subsequently, we formulate a numbering strategy based on double-step lengths to characterize the point cloud positions and determine the dense point cloud regions. Finally, we propose a random oscillations algorithm to generate additional points to accomplish the workspace improvement. Furthermore, we concludes with simulation to demonstrate that the point cloud distribution in the improved workspace is more uniform, and there is also a significant improvement in computational efficiency, compared to traditional generation algorithms.