Development of a practical photogrammetric network design using evolutionary computing

Photogrammetric network design is the process of optimising a network configuration in terms of the accuracy of object points. This accuracy is improved by selecting the best set of intersection rays in a bundle adjustment. This paper explains the approach followed to achieve the development of a practical network design using a real robotic system. The aim is to integrate a simulation-based network design into a practical vision metrology system. The simulation is carried out with a population search metaheuristic inspired by the evolutionary computing paradigm. The proposed approach uses an analytical criterion to search for a suitable first-order design in a fraction of the computational time required by the evolutionary approach using a rigorous bundle adjustment. The combination of an evolutionary algorithm simulation with the very well-established bundle adjustment provides a straightforward approach to network design. An example of a photogrammetric network design is provided to illustrate the result of a complete network design.