Algorithms for Fitting Cylindrical Objects to Sparse Range Point Clouds for Rapid Workspace Modeling

Current methods for construction site modeling employ large, expensive laser range scanners that produce dense range point clouds of a scene from different perspectives. While useful for many purposes, this approach is not feasible for real-time applications, which would enable automated obstacle avoidance and semiautomated equipment control, and could improve both safety and productivity significantly. This paper presents human-assisted rapid environmental modeling algorithms for construction, and focuses on cylindrical object fitting algorithms. The presented algorithms address construction site material of cylindrical shape. Experiments were conducted to determine: (1) the effect of the ratio of length to diameter of the cylinder to the accuracy of the results, (2) the effect of the angle of view to the accuracy of the results, (3) the minimum number of scanned points required to give adequate modeling accuracy for cylinders of various length to diameter ratios. The results indicate that the proposed algorithms can model geometric primitives used in a construction site rapidly and with sufficient accuracy for automated obstacle avoidance and equipment control functions.