OPTILIGHT: Mathematical Optimizations for Human Centric Lighting

RESEARCH SUMMARY: The overall goal of this project is to make lighting control systems more centered towards the human user. This requires not only better insights in how humans experience light but also demands quantified models and optimization algorithms that are executed by automated lighting control systems. Despite the growing scientific understanding of the impact of light on, for instance, wellbeing, performance, circadian rhythms and sleep, benefits of this understanding cannot (yet) easily be harvested in practical systems. We lack scalable algorithms that can be used in automated systems and that can be deployed in different environments without extensive tuning by experienced lighting experts. Scalability towards broad deployment is a key sub goal of this project. Humans want to experience light as a natural given. Having to adjust the light setting regularly is not attractive. Moreover, people are usually not aware of the longer-term effects of light so they don?t not necessarily select the optimal light setting. On the other hand, automatic controls often fail to offer a comfortable and unobtrusive natural experience and even tend to irritate people. Hence, there exists a huge gap between results obtained in controlled environments and practical deployment. Although control theory and optimizations using statistical signal processing are well established areas, to date they not widely used for lighting control. Yet, in audio and video processing and in gaming, models on human factors are successfully being used by automated systems. Hence we have reasons to be confident that better models human experience and perception can improve automatic control systems. However, it is not straightforward to capture human experience in equations. We believe that an important step is to better quantify the reliability of such expressions and to take this into account in probabilistic algorithms. The project has four Work Packages (Fig 4). WP1 and WP 2 focus on obtaining better quantified models on human perception and experience. WP1 focusses on controlled experiments while WP2 takes a big data approach and extracts statistical models from observing users in natural environments. WP3 uses the quantified models and reliability information in formal optimizations. We plan to apply the principles of Statistical Signal Processing to Lighting Control and to account for incomplete knowledge that the system has about the users and learn from (hopefully) very sparse user interventions in the lighting conditions. WP4 addresses how to use this in self configuring, rapidly learning smart lighting control infrastructures. UTILISATION SUMMARY: Light plays an important role in our lives. First of all, light affects people on a functional level by influencing the visibility of objects and people?s performance of visual tasks (e.g., Boyce, 2014). How well light conditions are chosen can significantly enhance or deteriorate the effectiveness of office workers (R. Cooper, 2008). Light also has a psychological effect by influencing how people feel. For instance, light can influence the subjective impression of a room (Flynn, 1977; Vogels, 2008) and people?s mood (Knez, 1995; McCloughan et al., 1999). Finally, light has an impact on biological mechanisms. Light regulates the human circadian rhythm (Duffy & Wright, 2005). Light also induces vitality and alertness (Cajochen, 2007; Smolders, de Kort & Cluitmans, 2012, 2015; Smolders, de Kort & Van den Berg, 2013). Moreover, appropriate light settings can improve sleep and avoid disorders such as Seasonal Affective Disorder (SAD) (Terman & Terman, 2005). Through these slow processes, light conditions have significant effects on wellbeing and productiveness. But ? if offered incorrectly - light itself can also be a stressor. It can cause discomfort, dissatisfaction and it can deteriorate motivation and performance (Berman et al. 1994; Rea et al. 1985; Boyce, 2014; Veitch et al. 2008). The LED as a light source, thus as a technical means to "spray Lumens", is reaching technological maturity. This paves the way for a next wave of innovation in how we experience light and how we can dynamically adapt our environment to improve our wellbeing. LEDs allow a finer granularity of lighting control. Not only the intensity but also other parameters such as color temperature, beam size and beam shape can be set and adapted during the day. Thereby LEDs may enable significant improvements to health and wellbeing. However, in practice it turns out to be very difficult to capture these benefits. The effort to translate our knowledge about the impact of light on humans into new installations and new environments is still prohibitive and requires a lot of designs by experts and custom adaptations. In order to effectively valorize the outcome of the project and to ensure a transfer our results to state-of-the-art systems (see section 5.1 for our plan) the project is conducted in the context of an existing strong bilateral cooperation between TU/e and Philips in the Intelligent Lighting Institute (ILI) (see chapter 5). Timeliness and Urgency: We believe that this research is urgent. Conventional lighting is rapidly being replaced by LED lighting that lasts for 50,000 hours of active on-time. Unless optimized behavior can be added to new systems in a routine fashion, this replacement will remain limited to replacing legacy by LED. So within a few years a great window of opportunity will largely have evaporated. In parallel, over the coming decades depression and burnout will be among the primary personal, societal, health-related and economic burdens in developed countries.