HELSINKI, March 25 (Xinhua) — What would it be like when artificial intelligence (AI) is used for building houses?
A total of 230 researchers from 30 countries and regions studied the possibilities of applying AI in architecture, engineering, and construction at an online conference arranged by the Finnish Association of Civil Engineers on Thursday.
The two-day event was reported as “the first of its kind in the world”, with themes ranging from smart building, data analytics to impactful AI.
Researcher Laura Pellegrini from Politecnico di Milano, Italy, took up AI solutions for the post-COVID-19 conditions where mandatory social distancing measures are causing large delays when entering and leaving buildings through elevators and stairs. Due to rapidly-changing safety measures, decision-making processes based on building databases may be unreliable.
She proposed applications for collecting occupancy data enabling the analysis of occupancy patterns. It would support a safe transition towards more flexible post-pandemic workplaces with fast-changing requirements, highly variable occupancy levels, and real-time indoor conditions adjustments.
Researcher Raghuram Kalyanam from the Technical University of Kaiserslautern, Germany, introduced a re-enforcement learning (RL) system that would make glare-reducing electrochromic (EC) window glass serve people better. The RL agent is rewarded for providing visual comfort to occupants by choosing the right electricity combinations of EC glass or penalized for bad combinations.
Dr. Chankyu Lee from the University of Florida, U.S., took up AI as an answer to the challenges of the current free-form architecture.
Complex free-form structures are trending in contemporary architecture. But it’s not an easy task to design and construct irregularly shaped buildings, of which aesthetics, statics, scale, and manufacturing technologies are quite different from ordinary ones.
Lee said as the processes of free-form architecture are more complicated and high-priced, the balance between design intent and cost efficiency becomes crucial. He suggested a rationalization method as a means to efficiently classify and penalize a free-form surface by utilizing generative and parametric designs. Enditem