Research works aims to propose a flexible method for balancing crankshafts. Nowadays, the crankshaft unbalance measurement is performed using balancing machines, which are highly dependent on the geometry and on the typology of crankshafts. Then, it seems worthwhile to replace the mechanical measuring equipment by a contactless measurement system for digitizing the surface geometry of the part. This would allow a flexible measurement of the mechanical parts. In this context, we highlighted two major problems: the calculation of unbalance from noisy and incomplete digitized data, and the choice of acceptable scanning systems in relation to required balance quality. The second problem also includes the development of a path planning strategy for 3D scanning of mechanical parts which respects the demands of quality and cycle time of production line. A method for unbalance calculation based on voxels was developed. A protocol for qualifying digitizing sensors was also established. This one is based on quality indicators previously developed at our laboratory (LURPA). Some new quality indicators to take into account the geometric complexity of crankshafts were also proposed. The method of path planning is also based on voxelization, an adaptive voxelization, in which the voxel size is adjusted according to the industrial needs while maintaining the characteristics of the field of view of the sensor. The overall approach of balancing by measuring the external skin geometry of mechanical parts was validated by digitizing 7 crankshafts with two different 3D scanning systems.
