Professor Thanos Tzempelikos and his team use a programmable high dynamic range image sensor that automatically captures and processes real-time luminance images to correlate lighting distributions with behavioral responses.

Daylight and window views positively affect health and well-being in many ways. However, children with Autism Spectrum Disorder (ASD) are hypersensitive to environmental stimulation, which leads many educational facilities to use windowless spaces or fully block natural light to avoid potential distractions.

Thanos Tzempelikos, Professor at the Lyles School of Civil and Construction Engineering and Ray W. Herrick Labs faculty, leads a $600,000 collaborative research project with Ball State University, sponsored by National Science Foundation, to solve this problem.

“To realize our vision, we formed a multidisciplinary team with expertise on daylight assessment, control, engineering of indoor environments, autism behavior, and the psychology of interior design,” says Tzempelikos.

This interdisciplinary research includes an experimental data-collection design process that includes a sensing, observation, and communication framework, designed to measure and record the daylight variables and children's behavioral responses in a systematic, synchronized way.

Experimental design. This project will first discover evidence linking daylight exposure (with different window shading options) and behavioral responses of ASD children with ASD, through extensive experiments. The experiments are performed with up to 50 children, 5-10 years old, with confirmed ASD diagnosis and normal or correctable vision, in collaboration with several autism facilities in Indiana. The tests are conducted at the Health and Environmental Design Research Lab at Ball State University, directed by Dr. Shireen Kanakri (Co-PI of the project).

A space with south-facing windows, providing natural light at all times of the day, was retrofitted for the purpose of the experiments, adding child-friendly features and a play area. The amount and distribution of transmitted daylight will be controlled by adjusting interior roller shades and their optical properties (openness factor and total visible transmittance). These have a combined nonlinear effect on indoor daylight conditions and glare control. In this way, researchers will be able to cover variable daylight luminance distributions, which will also naturally vary with time and sky conditions. The test space connects with a control room through a one-way observation window.

Behavioral measurements. Before the experiments, researchers will establish general baseline data about each child using the Adaptive Behavior Scales and Sensory Integration Assessment (SI). This data will be used to assess how each child copes with the daily sensory information load.

The following seven behaviors will be monitored during the experiments with variable daylighting distributions: repetitive movement, repetitive speech, ear covering, hitting response, loud sounds, blinking eyes, and noncompliance to requests. Real-time behavioral data, synchronized with each other and with environmental (daylight) measurements, will be recorded using a professional system event recorder.

A programmable digital camera synchronizes the saved video with accelerometer data streams. Timing and type of behaviors will also be coded offline by two independent raters using a custom video coding software application. In addition, every 10 seconds, a video recorder linked to the lighting measurements will record all activities and sound levels (to isolate other factors that can affect behavior) during the observation period. At the end of each day of observation, semi-structured interviews with parents/caregivers will provide any additional information.

Daylight measurements. The Luminance distribution across the entire visual scene will be monitored using a programmable High Dynamic Range Imaging (HDRI) sensor with a wide-angle fisheye lens, mounted on a wall behind the child to cover the subject’s field of view and beyond. Additional cameras will capture the window and monitor the shading status at all times.

The HDR images are processed using Python scripts to perform per-pixel luminance analysis. This sensing and computational module is a powerful tool for image processing and for monitoring real-time luminance distributions, providing the following general lighting metrics: average and maximum scene luminance; contrast ratio between the task area and the window; luminance uniformity ratio; and visual comfort metrics such as daylight glare probability.

Experimental phases. The experiments will be conducted in four phases: pilot testing, baseline data, main experiment, and post-experiment. All observations will be conducted individually for each child, who will be tested four times to capture different daylight distributions and check the reliability of the data. The experiments will run for up to 48 days (7 weeks).

Each child will choose a task (puzzles, coloring, playing in the sandbox, or reading a story) during each visit. During each phase, the 7 behaviors will be collected every 5 seconds with synchronized monitoring of the daylight measurements. To explore correlations between daylight and children’s responses, researchers will vary indoor daylight distributions in a randomized order (by adjusting shading position and optical properties) during the main experimental phase and simultaneously collect behavioral data, highlighting any changes in children's sensitivity and sensory behavior. All physical and observational data are automatically recorded and synchronized in a database labeled by participant number, visit, and other relevant information.

Connecting daylight metrics with behavioral responses using computational techniques. Using the collected datasets, statistical and machine-learning-based computational models will be developed to predict dynamic behavioral relationships to daylight-induced stimulation and provide insights into response classification. Finally, daylight and reverse-engineering modeling will translate the findings into new design and operational guidelines and best daylighting practices. These will be implemented and evaluated in real autism schools, in collaboration with teachers, parents, and autism community stakeholders.

Challenges. When asked how the research team will ensure that experiments and findings are representative and applicable across different ASD populations and environments— Professor Tzempelikos shares, “That is not possible in any studies involving human subjects, especially of this nature. However, we hope to discover patterns of similar behavioral responses to daylight in different groups of children, which will lead to clustered response classification and generalization of results. More evidence will be provided at the last phase of the project, during the full implementation of the findings in autism schools with larger groups, where we will evaluate evidence-based best daylighting practices in autism schools and refine models as needed.”

Researchers studying how daylight affects children with autism spectrum disorder (ASD) face two major challenges: collecting reliable data and accurately interpreting it. Children with ASD are often sensitive to their environment, making it difficult to isolate the effects of daylight from other influences. Distractions and varying responses complicate data collection. To address these issues, researchers use carefully designed experiments, provide support and breaks for participants, and employ advanced analytical techniques—such as Bayesian statistics combined with machine learning—to better understand the subtle ways daylight impacts behavior.

Impact. The research aims to improve physical and mental health outcomes by tailoring natural light environments to the needs of children with ASD. This includes developing engineering solutions for optimal daylight exposure and providing guidance to parents, teachers, designers, and physicians. The overarching goal is to foster healthier behaviors, emotional regulation, and social development by addressing issues such as disrupted sleep and misaligned circadian rhythms through informed daylight management.