In smart lighting systems, a controller interprets lighting data collected by sensors and adjusts luminaire outputs or shade positions in order to ensure visual comfort of occupants and minimize energy use. The design of such systems requires close coordination between architectural, lighting, and controls designers. Effective computational design methods are essential to help designers manage the complexities of smart lighting design and realize the full potential of smart lighting technology.
This PhD thesis investigates three related problems in current smart lighting system design workflows. First, data exchange between designers is limited. The same data may need to be entered multiple times, and design changes are time-consuming. Based on existing work on space data models, views are defined for architectural, lighting, and controls design. These views facilitate automated data exchange between design domains. Objects in each view have rich property sets that are relevant for smart lighting systems. Second, lighting as well as lighting controls design require considerable domain knowledge . At the same time, they involve highly repetitive tasks that are currently executed largely manually. An example is the placement of sensors and actuators. Existing luminaire placement methods are limited to individual spaces with simple geometries and window arrangements. We will therefore investigate methods to automate placing these objects using both procedural modeling and optimization methods [2,3]. Third, the quality of lighting simulation in existing systems is limited. Indirect illumination is rarely taken into account. In this work, we will use state-of-the-art lighting simulation tools and material definitions to obtain a much more accurate simulation of the resulting lighting . To keep simulations interactive, the challenge will be to identify reusable computations for different lighting situations. This will also allow an automatic optimization for given lighting goals.