Abstract: For the measurement problem of the 3D long and thin bent tube, an autonomous measurement system consisting of the industrial robot and multiple sensors is presented. An optimized scan path planning algorithm is proposed based on genetic algorithm, to autonomously scan the point cloud of the tube (as a crucial part of the measurement). With the algorithm a collision free scan path is acquired, satisfying that all the segments of the tube can be scanned, and the scan times are as small as possible. And the robot pose at each key scan point is restricted in a certain range around the vertical direction. Moreover, the searching step is updated based on the forward cover coefficient after the generation of each population, and the sampling interval of the new sample individual is also adjusted for quick acquisition of better individuals. In the part of the performance verification of the scan path planning algorithm, the simulation on a bent tube is implemented firstly. Then the scan path planning simulations of cylinder segment and toroidal segment, which are placed with different pose, are conducted. Finally, the simulation and the practical experiment of the scan path planning of a practical car tube are carried out. As the simulation results, the scan length of the best individual increases with the number of iterations, and the coverage ratios of the two kinds of tube segments and the car tube are all above 0.99. The results demonstrate that the number of the planned key scan points is small and the scan volumes of the key scan points can almost scan the tube completely. Simulation and practical experiment results indicate that the algorithm can obtain a suitable scan path for complex tubes.