Aladdin: Reimagining Design with AI
|
Artificial intelligence (AI) is changing the way people work. As a manifestation of human ingenuity, design is a vital testbed for studying how we may meet the opportunities and challenges posed by AI. Based on the fusion of CAD and CAE, Aladdin is an experimental platform for reimagining design in the coming era of AI. The power of Aladdin derives from two different sources: generative design and machine learning.
Generative design is a tabula rasa methodology inspired by biological evolution. Once the design criteria and constraints of a product are specified, a piece of generative design software uses evolutionary computation to efficiently explore the entire parameter space supported by the software to find optimal solutions. During the iterative search for feasible solutions, the software automatically constructs a vast number of forms at each step, tests their functions using numerical simulations, evaluates their quality based on the given criteria and constraints, and then selects those that are closer to the goal for the next step. By repeating these computational routines many times, a variety of designs that meet the goal gradually emerge. Engineers then review these outputs, often with the aid of interactive visual analytics for intuitive evaluation and comparison across the board, and choose one or more designs for prototyping. This paradigm shift in design methods entails a fundamental change of mindset for design thinking that must be addressed in the engineering education of future workforce.
Machine learning, on the other hand, collects and analyzes tabula inscripta data from human designers with the goal to translate their intelligence into computational models so that their creativity can be preserved and reproduced. Machine learning recognizes the fact that design is not only to meet the needs of human but also driven by the intuition of human and there are simply no well-defined laws capable of predicting human behavior.
|
In Arabic folklore, Aladdin is a tale that features a magic lamp able to generate whatever its owner wants. In the field of design, AI has somewhat realized the fairytale — designers only need to specify what they want and AI would bring their wishes to life.
Software
Tutorials
Educational Activities
Engineering Domains
Published Applications
Most design problems have multiple objectives. A multi-objective AI agent outputs a set of solutions along the Pareto frontier,
where it is impossible to improve the performance of a solution on one objective without compromising the performance on another.
The Pareto frontier is where AI reaches its limit. Beyond this point, human judgement is needed to decide on the final solution to complete the design process.
This project is supported by the National Science Foundation (NSF) under grant numbers #1918847 and #2105695. Any opinions, findings, and conclusions or recommendations expressed in this material, however, are those of the authors and do not necessarily reflect the views of NSF.
|
