Robust, precise and scalable: A phase-encoded pattern for visual X, Y, Θ positioning*

摘要

At small scales, vision is a widely used contactless sensor allowing multi-dimensional measurements, notably position, orientation, displacements and relative distances. The most convenient procedure consists in processing natural features present in images of the scene of interest. However, this approach results in performances that are object-dependent, sensitive to noise and restricted to measurement ranges smaller than the field of observation. Different kinds of structured patterns attached to the moving object of interest were proposed to overcome these limitations. This paper presents a phase-based visual X, Y, $Theta$ positioning method combining nanometric resolution, large measurement range and robustness against common types of images alterations. The paper describes the phase-based position coding/decoding principles and then demonstrates the large range-to-resolution ratio as well as the method capabilities to reconstruct the correct position despite significant levels of image noise/occlusion. Finally, the trade-off between image size, resolution obtained and computation speed is discussed.