Towards a generic BRDF/BTF measurement system: Improving visual realism in robot simulators using robots

摘要

It is well understood that the Bidirectional Reflectance Distribution Function (BRDF) and the Bidirectional Texture Function (BTF) are the most adequate description of real world surface appearances. However, their use is still marginal in robotics simulators, even though real-time simulation of visual sensors is common nowadays. For visual realism two aspects of the simulated objects are important: geometric and surface properties. Assuming a sufficiently accurate geometric model is available, through CAD modeling or a 3D scan, its surface or reflectance properties have to be defined. This paper describes a generic measurement system to acquire a BRDF/BTF as basis for an analytic reflectance model that can be used for simulation of visual sensors in real-time robotics simulators. We designed the system and its architecture under the constraints that it is easy to setup, does not permanently need storage space and only requires components that are already available in most robotic labs, while providing sufficiently accurate results. Our goal is to lower the entry barrier for roboticists to the extent that they can benefit in practice from the progress in BRDF/BTF research.