Position uncertainty in space scene reconstruction for simulation-based lighting control

Dynamic, three-dimensional models of existing facilities that are updated automatically based on data provided by advanced sensing technologies appear increasingly feasible. Simulation-based lighting control is one potential application domain, where, besides information regarding external obstructions and sky conditions, accurate, up-to-date models of space boundaries, openings, and, possibly, furniture and light fixtures are required to support controller decisionmaking. In this paper, we describe a procedure that uses spatial reasoning and constraints to support the reconstruction of space scene models that are sufficient for lighting simulation. Specifically, the procedure addresses the problem of position uncertainty in sensor data. Its potential is demonstrated in a test setting. Daylight performance indicators are computed for an as-built model of a test space and compared with those obtained for a model generated with the procedure.