“`html
Optimization through artificial intelligence presents numerous advantages for mechanical engineers, such as quicker and more precise designs and simulations, enhanced productivity, decreased development expenses via process automation, and superior predictive maintenance alongside quality assurance.
“When individuals consider mechanical engineering, they envision fundamental mechanical instruments like hammers and … devices such as cars, robots, cranes, yet mechanical engineering encompasses a vast range,” remarks Faez Ahmed, the Doherty Chair in Ocean Utilization and associate professor of mechanical engineering at MIT. “Within the realm of mechanical engineering, machine learning, AI, and optimization are significantly influential.”
In Ahmed’s class, 2.155/156 (AI and Machine Learning for Engineering Design), students apply tools and strategies from artificial intelligence and machine learning to mechanical engineering design, concentrating on the development of innovative products and confronting engineering design obstacles.
“There’s ample justification for mechanical engineers to consider machine learning and AI to effectively hasten the design process,” states Lyle Regenwetter, a teaching assistant for the course and a PhD candidate in Ahmed’s Design Computation and Digital Engineering Lab (DeCoDE), where the research is centered on formulating new machine learning and optimization techniques to analyze intricate engineering design issues.
Initially offered in 2021, the course has swiftly risen to be one of the Department of Mechanical Engineering (MechE)’s most sought-after non-core courses, drawing students from various departments across the Institute, including mechanical and civil and environmental engineering, aeronautics and astronautics, the MIT Sloan School of Management, as well as nuclear and computer science, along with cross-registered students from Harvard University and other institutions.
The course, accessible to both undergraduates and graduates, emphasizes the execution of advanced machine learning and optimization approaches in the framework of tangible mechanical design challenges. From constructing bike frames to urban grids, students engage in competitions related to AI for physical systems and address optimization hurdles in a classroom environment powered by friendly rivalry.
Pupils are presented with challenge problems and initial code that “provided a solution, but [not] the optimal solution …” elaborates Ilan Moyer, a graduate student in MechE. “Our objective was to [figure out], how can we enhance this?” Live leaderboards motivate students to consistently improve their techniques.
Em Lauber, a system design and management graduate student, shares that the experience allowed for discovering the application of what students learned and the practical skill of “literally how to code it.”
The syllabus includes discussions on scholarly papers, and students also undertake practical exercises in machine learning geared towards specific engineering challenges, such as robotics, aircraft, structures, and metamaterials. For their final project, students collaborate on a team endeavor that applies AI methods to address a complex issue of their choosing.
“It is delightful to see the varied scope and high caliber of class projects,” notes Ahmed. “Student initiatives from this course frequently result in research publications and have even garnered awards.” He highlights a recent paper titled “GenCAD-Self-Repairing,” which won the American Society of Mechanical Engineers Systems Engineering, Information, and Knowledge Management 2025 Best Paper Award.
“The most rewarding aspect of the final project was that it granted every student the chance to implement what they’d learned in the class to a field that fascinates them,” asserts Malia Smith, a graduate student in MechE. Her project focused on “marked motion captured data” and examined predicting ground force for runners, an endeavor she described as “truly fulfilling” due to its unexpectedly impressive results.
Lauber utilized the model of a “cat tree” design with diverse modules of poles, platforms, and ramps to develop personalized solutions for specific cat households, while Moyer devised software that is creating a novel type of 3D printer architecture.
“When you observe machine learning in popular media, it appears highly abstracted, giving the impression that something extremely intricate is occurring,” remarks Moyer. “This class has unveiled the reality.”
“`