Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

University of Calgary

$32,000.00

Automated Design Assistants for Human-AI Collaborative Design

EF 2024-25

Résumé
The human design process is a ubiquitous element of modern society, playing a critical role in the technologies we live in, impacting countless industries such as aerospace, automobile, shipping, agriculture, etc. Augmenting the design process through AI and automation can reduce cost and increase productivity, improving global prosperity. The partnership between humans and AI technologies has a significant potential to extend the ability of humans to address complex design problems. My research mission is to revolutionize the modern engineering design process through human-AI partnerships and design assistants. The long-term objective of my research program is to augment engineering design through human-AI teams by coalescing the complementary strengths of humans and AI. Towards this overarching objective, I focus on three short-term objectives in the first five years: (a) Quantify the effect of priors, strategies, and information exchange behavior on the performance of human designers and design teams in engineering design, (b) Create AI-based design assistants using data and observations from human experiments, and (c) Quantify the effect of AI tools on the design behavior and strategies of human designers. A novelty of my program is to create sample efficient reinforcement learning (RL) agents for design automation learning from humans. RL has been demonstrated to create agents with super-human performance in fields such as robotics and gaming. However, challenges such as expensive physics models and lack of generalization limit the application of RL in engineering design. Curiosity-based RL algorithms require millions of actions to train. Unlike RL, data shows that even a novice human designer can generalize across several design tasks in a few thousand actions. I will use human subject experiments conducted in a web-based virtual design studio to collect training data. I will use recent advances in representation learning and ablation studies to quantify the effect of various human priors and heuristics on design performance. I will use the Skill-Transfer framework and imitation learning algorithms to create sample-efficient RL agents that learn from human data for trade space exploration in engineering design. Meta-learning and transfer learning will enable the generalization of the results to a broad spectrum of engineering design tasks. For synergistic integration of AI tools and humans, my program evaluates the effect of AI assistants on human design teams. Designer behavior and performance in baseline (control group without AI assistants) and augmented (with AI tools and intervention) will quantify the effect of AI assistants on the performance and design behavior of human design teams. The outcomes include heuristics and priors for RL agents, AI tools (intelligent design assistants), and new approaches (design methods) for human-AI collaboration.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Alberta, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
RGPIN-2023-04282
Name nom
Thekinen, Joseph
Department département
Department of Mechanical and Manufacturing Engineering
Organization id
11
Institution établissement
University of Calgary
Province en
Alberta
Province fr
Alberta
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
Award amount
32000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1512
Committee name en
Mechanical Engineering
Committee name fr
Génie mécanique
Area of application code
1206
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Engineering
Area of application fr
Génie
Research subject code
2100
Research subject group en
Mechanical engineering
Research subject group fr
Génie mécanique
Research subject en
Mechanical engineering
Research subject fr
Génie mécanique
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Automated Design Assistants for Human-AI Collaborative Design
Keywords
mechanical engineering design; design automation; intelligent design assistants; human-ai collaboration; human subject experiments
Application summary
The human design process is a ubiquitous element of modern society, playing a critical role in the technologies we live in, impacting countless industries such as aerospace, automobile, shipping, agriculture, etc. Augmenting the design process through AI and automation can reduce cost and increase productivity, improving global prosperity. The partnership between humans and AI technologies has a significant potential to extend the ability of humans to address complex design problems. My research mission is to revolutionize the modern engineering design process through human-AI partnerships and design assistants. The long-term objective of my research program is to augment engineering design through human-AI teams by coalescing the complementary strengths of humans and AI. Towards this overarching objective, I focus on three short-term objectives in the first five years: (a) Quantify the effect of priors, strategies, and information exchange behavior on the performance of human designers and design teams in engineering design, (b) Create AI-based design assistants using data and observations from human experiments, and (c) Quantify the effect of AI tools on the design behavior and strategies of human designers. A novelty of my program is to create sample efficient reinforcement learning (RL) agents for design automation learning from humans. RL has been demonstrated to create agents with super-human performance in fields such as robotics and gaming. However, challenges such as expensive physics models and lack of generalization limit the application of RL in engineering design. Curiosity-based RL algorithms require millions of actions to train. Unlike RL, data shows that even a novice human designer can generalize across several design tasks in a few thousand actions. I will use human subject experiments conducted in a web-based virtual design studio to collect training data. I will use recent advances in representation learning and ablation studies to quantify the effect of various human priors and heuristics on design performance. I will use the Skill-Transfer framework and imitation learning algorithms to create sample-efficient RL agents that learn from human data for trade space exploration in engineering design. Meta-learning and transfer learning will enable the generalization of the results to a broad spectrum of engineering design tasks. For synergistic integration of AI tools and humans, my program evaluates the effect of AI assistants on human design teams. Designer behavior and performance in baseline (control group without AI assistants) and augmented (with AI tools and intervention) will quantify the effect of AI assistants on the performance and design behavior of human design teams. The outcomes include heuristics and priors for RL agents, AI tools (intelligent design assistants), and new approaches (design methods) for human-AI collaboration.