Design by Machine Learning (Deep Learning)
(i) Design Automation by Deep Learning
Recent advances in deep learning enable machines to learn existing designs by themselves and to create new designs. Generative adversarial networks (GANs) are widely used to generate new images and data by unsupervised learning.
This research proposes a design automation process by combining GANs and topology optimization. The suggested process has been applied to the wheel design of automobiles and has shown that an aesthetically superior and technically meaningful design can be automatically generated without human interventions.
(ii) Eliciting Styling Preference on Vehicle Design by Machine Learning
Although eliciting consumer preference is a core part in design, capturing preferences on form (styling) is still a challenge. In particular, we have been examining the trade-off between forms and functions in a vehicle. We elicit consumers’ preferences on product design using state-of-the-art Machine Learning and parameterized 3D vehicle shapes. This research estimates willingness-to-pay for specific design elements.
(iii) Forecasting Demand of Hybrid Trucks by Hierarchical Bayesian
The market for delivery services has grown substantially along with the development of e-commerce. In an effort to reduce pollution caused by delivery trucks due to low fuel efficiency and high emissions, the Korean government is considering converting the diesel-powered delivery trucks in the current market to hybrid trucks. The objective of this research is to predict truck drivers’ demand for converting to hybrid trucks depending on converting costs and increase in fuel efficiency by using Machine Learning.
Supported by KOTI(The Korea Transport Institute), University of Michigan Transportation Research Institute(UMTRI), Rackham Graduate Student Research Grant
- Design Automation by Integrating Generative Adversarial Networks and Topology Optimization, IDETC, 2018. (PDF)
- Form + Function: Optimizing Aesthetic Product Design via Adaptive, Geometrized Preference Elicitation, Marketing Science (under review). (PDF)
- Influence of Automobile Seat Form and Comfort Rating on Willingness to Pay, International Journal of Vehicle Design, 2017. (PDF)
 Autonomous Vehicle Sharing System Design
(i) Sharing Economy of Autonomous Electric Vehicles
Car sharing services promise “green” transportation systems. Two vehicle technologies offer marketable, sustainable sharing: autonomous vehicles (AVs) eliminate customer requirements for car pick-up and return, and battery electric vehicles entail zero emissions. Designing an autonomous electric vehicle (AEV) fleet must account for (1) the relationships among fleet operations, (2) charging station (CS) operations, (3) electric powertrain performance, and (4) consumer demand. This research proposes a system design optimization framework integrating these four subsystem problems. This framework is used to examine AEV sharing system profitability and feasibility for a variety of market scenarios. The results provide practical insights for service system decision makers.
(ii) Autonomous Taxi Service Design
When it comes to autonomous vehicles, most manufactures focus on the technological side and not on the design of how to provide the service. When traditional taxis are replaced by autonomous ones, there will be substantial behavioral changes by customers. The service providers will only succeed if they develop an optimal UI/UX design that adjust to these changes. This research develops a guideline of autonomous taxi service from the customers’ perspective, and further provides design requirements of autonomous vehicle development for manufacturers.
Supported by CRC(Convergence Research Center), Kakao Mobility, NRF (National Research Foundation of Korea), RED&B(KAIST), Dow Distinguished Award for Interdisciplinary Sustainability, MCube Grant
- Autonomous Electric Vehicle Sharing System Design, Journal of Mechanical Design, 2017. (PDF)
 Electric Vehicle Design for Market Systems
We have been working on electric vehicle (EV) design problems by integrating electric powertrain simulation, charging station location network, consumer demand, and public policy models, into a large decision making framework. This research compares the public investment impacts on EV markets in China and US. In addition, We are exploring how to make design decisions today by taking future long-term design evolution into account under engineering and marketing uncertainties. We adopt real option theory used in Financial Engineering and RBDO (Reliability-Based Design Optimization) to hedge design risks and create smooth design transitions.
Supported by Toyota Motor Corporation, Automotive Research Center (US Army), Graham Doctoral Fellowship
- Reliability-based Design Optimization (RBDO) for Electric Vehicle Market Systems, IDETC, 2017. (PDF)
- A Real Options Approach to Hybrid Electric Vehicle Architecture Design for Flexibility, Journal of Mechanical Design, 2018. (PDF)
- Public Investment and Electric Vehicle Design: A Model-based Market Analysis Framework with Application to a USA-China Comparison Study, Design Science, 2016. (PDF)
- An Integrated Design Approach for Evaluating the Effectiveness and Cost of a Fleet, Journal of Defense Modeling and Simulation, 2016. (PDF)
- Designing Profitable Joint Service-Product Channels, Journal of Services Marketing (under review). (PDF)
- Integrated Decision Making in Electric Vehicle and Charging Station Location Network Design, Journal of Mechanical Design, 2015. (PDF)
- A Framework for Enterprise-driven Product Service Systems Design, ICED, 2013. (PDF)
- Integrated Design Process of Conjoint Analysis and TRIZ, KSIME, 2009. (PDF)
- Integration of Marketing Domain and R&D Domain in NPD Design Process, Industrial Management & Data Systems, 2007. (PDF)
 Multidisciplinary Design Optimization
An engineering system problem (e.g., vehicle design) may not be solvable due to the size and complexity of models. Analytical Target Cascading (ATC) method has been developed to address this issue by partitioning a system into subsystems and coordinating them to meet a system goal. We have been designing commercial vehicle systems and hybrid vehicle architectures using ATC. However, we realized that ATC with parallel solving approach often fails to achieve convergence or involves high computational cost. Therefore, we research on robust convergence strategies for complex industrial applications that should be solved in parallel.
Supported by Hyundai Motor Company, Altair University Fellowship
- Modified Augmented Lagrangian Coordination and Alternating Direction Method of Multipliers with Parallelization in Non-hierarchical Analytical Target Cascading, Structural and Multidisciplinary Optimization, 2018. (PDF)
- Decomposition Based Design Optimization of Hybrid Electric Powertrain Architectures: Simultaneous Configuration and Sizing Design, Journal of Mechanical Design, 2016. (PDF)
- Solving Multiobjective Optimization Problem Using Quasi-separable MDO Formulations and Analytical Target Cascading, Structural and Multidisciplinary Optimization, 2014. (PDF)
- Optimal Design of Commercial Vehicle Systems Using Analytical Target Cascading, Structural and Multidisciplinary Optimization, 2014. (PDF)