A Decision Support Software Tool for Reasoning about the Subjective Impressions of a Lighting Installation

The discipline of architecture is concerned with finding a balance between both the functional and the subjective aspects of a building environment. This involves managing contradictory requirements that are often difficult to resolve through purely numerical analysis; an example of this is an electrical lighting installation designed to evoke a desired subjective impression or ‘atmosphere’, which may conflict with the visual requirements for accurate or safe task performance. Despite this, few software tools exist that directly support an architect when dealing with information relating to the non-visual effects of lighting. A fundamental limitation in standard software tools is the reliance on numerical approaches for representing and reasoning about lighting and construction related information. In particular, when information is uncertain or completely unavailable, numerical formulae can be awkward or impossible to use in a reliable way. Work in the field of qualitative spatial reasoning has attempted to address these issues, and in this paper we present a prototype decision support software tool that reports on the subjective impressions of a lighting scheme, based on a qualitative spatial reasoning engine. Research in subjective response to lighting is reviewed and interpreted in the context of qualitative spatial reasoning, and the prototype system is compared to studies on subjective impressions. 1.0 INTRODUCTION The discipline of architecture is concerned with more than simply meeting practical criteria, such as: Can the building support the required load? Does the noise level, temperature, or airflow meet the appropriate health and safety standards? Architecture involves the study of how to direct a person’s perception of their environment, for example, to evoke a mood, or to convey an abstract concept. This involves managing possibly contradictory requirements that are often difficult to resolve through purely numerical analysis; an example of this is the subjective impression, or atmosphere of a space that can be evoked by lighting but which may or may not coincide with the visual requirements for safe and effective task performance. Despite the need to work with the subjective impressions that people experience, few software tools exist that directly support an architect when dealing with information relating to the non-visual effects of lighting. A fundamental limitation in standard software tools is the reliance on numerical approaches for representing and reasoning about lighting and construction related information. For example, the focus of many tools has been on providing computationally expensive simulations such as ray tracing to render or visualise an environment or to calculate luminance distributions across a space (such as (Ward 1994)). One problem is that the level of detail at which processing is being performed is often inappropriate, particularly for early stages of design. Furthermore, when information is uncertain or completely unavailable, numerical formulae can be awkward or impossible to use in a reliable way. For example, a lighting designer may be given preliminary building sketches where materials used and dimensions are only loosely described, and be required to produce a number of different lighting schemes that satisfy subjective and practical requirements. Other issues relate to usability. For example, computer simulations often result in a large amount of numerical data, involving a variety of units. The architect must then manually determine whether the desired aesthetic and functional requirements are being met, along with health and safety standards. A software tool is required that allows an architect to explore various lighting designs by quickly giving feedback on the non-visual effects of a lighting installation. Formalisms in the field of qualitative reasoning (a branch of artificial intelligence) have been developed that address the limitations raised by purely numerical systems (Bobrow 1984; Weld and de Kleer 1990; Kuipers 1994; Forbus 1996). The aim of qualitative reasoning is to identify and reason about coarse, qualitatively significant distinctions between object relations. It offers a more human-intuitive approach to working with information by relying on concepts such as causality, the nature of interaction, and by involving everyday terms that capture imprecision and vagueness automatically (such as very bright, fairly dim, compared to 356 lux). Qualitative spatial reasoning is a subfield that reasons about qualitative distinctions between spatial entities and relationships (Freksa 1991; Cohn and Hazarika 2001). In this paper we present a qualitative spatial reasoning engine that analyses an electrical lighting installation and reports on the subjective impressions that will be evoked. This prototype system is intended to assist an architect during the early stages of design by providing fast qualitative feedback on the subjective impact of a lighting installation. A framework for capturing qualitative relationships between subjective impressions and physical lighting configurations is presented that allows the compilation of a knowledge base used by the reasoning engine. 2.0 BRIEF REVIEW OF LIGHTING THEORY In standard lighting theory (Egan 1983; Sanders and McCormick 1993; CIBSE 1994; IESNA 2000; Bridger 2003) luminous flux is a measure of the light energy emitted by a light source, adjusted according to the eye’s response to certain wavelengths (for example wavelengths outside of the visible range are excluded). The units for luminous flux are lumens (lm) and can be calculated as: