Abstract: The structure and kinematics of a snake-shaped manipulator with 24 degrees of freedom are introduced, and the problems of the traditional Monte Carlo method in dealing with such hyper-redundant manipulator are analyzed. A point cloud contour extraction method based on k-nearest neighbours (k-NN) is incorporated into the traditional Monte Carlo method to solve the problems of uneven distribution of points, slow generation of point cloud and difficult contour extraction. In the end, the method is applied to the workspace determination of the snake-shaped manipulator, and its computational efficiency and practical value are verified.