Diversifying robotics with a one-of-a-kind collaborative teaching model
Toyota Research Institute and Amazon help fund the hybrid program that partners with Minority-Serving Institutions.
Michigan Engineering News
Experts
Chad Jenkins
Standing in front of a three-wheeled robot, a student holds up a panel from a maze-building kit, then pulls it away. The “MBot” follows. The student pushes the panel forward toward the machine. The MBot backs up.
This isn’t a particularly complex task for a robot. It’s merely maintaining a fixed distance from an object. But it has an important purpose for Chad Jenkins, a professor of robotics at U-M. He calls it the “Aha! moment” of his Robotics 102 course.
“The behavior you get looks really cool,” Jenkins said, “and it makes computation and robotics and engineering come to life.”
As the chief architect of Michigan Engineering’s unique Distributed Teaching Collaboratives initiative, Jenkins is committed to bringing epiphanies like this to students outside of Ann Arbor as well. The effort he’s leading aims to bridge gaps in both workforce needs for industry, and educational access for students from diverse backgrounds.
Video Transcript
Perry: We have some really bright hardworking students, and all they really need is the opportunity.
Chad: Now that we know how to do our wall follower. We know how to get code running on the robot. We’ve built the robot, now let’s talk about the software infrastructure.
Jana: We’re teaching Robotics102: Intro to Programing and AI at Morehouse College and also at Howard and Berea College.
Chad: To continue our work with what we call Distributed Teaching Collaborative, which is an open source approach to common teaching across universities to build stronger pathways for students into graduate school and success throughout their career in areas like robotics, A.I. and automation technology.
Todd: Faculty at smaller institutions, they’re more focused on providing support for basic computer science skills.
Jana: We’re really trying to make it easier for instructors to incorporate robot platforms into their curriculum, normally that’s something that takes a huge amount of design effort that most instructors don’t have the time to do.
Todd: So each student is paired with another student and they have their own robot. They get the instruction online, they interact with the Michigan TAs (Teacher Assistants) online.
Jana: Just within a couple of weeks, we have them coding real robot platforms and we have them driving these robots around to move autonomously.
Todd: The opportunity to take a class with a subject matter expert where students can go deep, or at least begin to go deep, has been very valuable for our students.
Chad: Through a collaborative teaching effort, we can build a bridge with our colleagues at other schools, but also a connection to other students who are going to create the innovations of the future. And so that we can see better diversity, better outcomes, more success in robotics.
Todd: We are introducing technology to students who may not have been able to interact with this technology any other way.
Perry: We know we’re going to make some mistakes. We know we’re going to hit some roadblocks, but we have the support across the collaborative to be able to help out.
Jana: We have some really excellent educators working with us. And so I think that their expertise and all of their new ideas keep making our courses here at Michigan better, also.
Chad: I think this is a model for how we can unleash more of the talent that’s out there, because we know that there are excellent people all across our nation, all across the state of Michigan, all across our world, that will do great if they have that opportunity.
Robotics is a rapidly growing field in the US. But robotics and AI programs can’t produce engineers fast enough to fill those jobs with American workers. At the same time, students from backgrounds that have historically been excluded from engineering remain at a disadvantage for getting those jobs too. They often don’t have access to the universities that are able to build out robotics programs quickly.
Through Distributed Teaching Collaboratives roboticists at U-M are connecting directly with minority-serving institutions (MSIs) to teach hybrid, collaborative courses. The benefits are multifaceted: students at Michigan interact with more diverse course cohorts, overstretched faculty at MSIs have partners for developing courses in a rapidly evolving field, and their students gain knowledge and contacts that can help them pursue advanced degrees.
“Our faculty colleagues at minority-serving institutions are teaching three to four courses per semester, leaving little time for the curricular innovations needed to prepare students for careers in robotics and AI,” said Jenkins.
The program started in 2020 as part of Robotics 101, the course that launched U-M’s one-of-a-kind undergraduate degree in robotics. Not only was Michigan Engineering the first top-ten institution to offer such a degree, its curriculum was designed to be inclusive from the start—to invite in rather than weed out.
Today the Distributed Teaching Collaboratives effort has grown to include four courses and five schools, including four Historically Black Colleges and Universities (HBCUs): Morehouse and Spelman Colleges of the Atlanta University Consortium, Berea College, Howard University, and Florida A&M University.
Support from the Sloan Foundation, Toyota Research Institute, and Amazon have helped make that possible.
“Our funders really helped us grow from just a couple of faculty working together into a larger, open-source model that invites faculty from any institution to engage with one another and serve more students across the world,” said Jenkins.
A bold pilot, supported by industry
The first course in the series, Robotics 101, flips the traditional order of engineering courses to give students an earlier taste of what the robotics field enables. It teaches linear algebra through robotics applications. Linear algebra is foundational to 21st-century computing, including AI, cryptography, and computer animation, as well as robotics. Conventionally, it follows calculus and is taught much later in math and engineering curricula. But calculus is not required to understand it. The approach puts students of all backgrounds on a more level playing field, whether or not they had the privilege of taking AP calculus in high school.
For the inaugural section in Fall 2020, Jessy Grizzle, the Elmer G. Gilbert Distinguished University Professor of Engineering at U-M, wrote a free digital textbook and team-taught with Dwayne Joseph, an assistant professor of mathematics and computational sciences at Morehouse. The hybrid class, a format they had been planning for even before the pandemic, included 39 U-M students, 12 at Morehouse and one at Spelman. They collaborated via Zoom, Piazza, and their own Discord groups, learning computational linear algebra, the open-source Julia programming language, and how to turn all of the math into effective algorithms. As of this semester, 57 students from the Atlanta University Consortium, 76 at Howard, 52 at Florida A&M, and nearly 1000 at Michigan have taken Robotics 101.
Grizzle notes that of Michigan students, a majority attend remotely. “The students taking it remotely from HBCUs are getting the same experience as the Michigan students,” he said. “They’re not getting a second-class education by being remote, as 80% of Michigan students are voting with their feet.”
In 2021, the team added Robotics 102, Introduction to AI and Programming. To teach it alongside Jenkins, they brought on board Jasmine Jones, an assistant professor of computer science, and Jan Pearce, an associate professor of mathematics, at Berea.
With a flipped classroom, students listen to lectures on their own time and work on programming robots in class at their own college or university—first enabling the robot to maintain a fixed distance from an object, then follow walls and eventually navigate a maze. The MBot robot kits used for the course cost $360 each, provided to MSI partners by U-M. This course was offered at Berea, Howard, the Atlanta University Consortium and Michigan in the last academic year.
In return for these resources, MSI instructors share their depth of knowledge as teachers, Jenkins says. The high course load means MSI faculty rapidly accumulate classroom experience. Likewise, students and faculty alike get a feel for the capabilities and experiences of those at partner institutions.
“Our Distribute Teaching Collaboratives pilot has provided personal interactions between students at Morehouse and Berea with Michigan, and these interactions have helped the Morehouse and Berea students to understand that they are performing just as well as the Michigan students, and that their struggles with the material are the same as the Michigan students,” Jenkins reported in 2022. “This creates the confidence needed to succeed and thrive at the graduate level.”
The Toyota Research Institute (TRI) supported the development of these first two courses.
“At Toyota, we recognize the importance of hiring a diverse set of roboticists to serve the global market for this emerging field. To do that, we need to invest in educational innovations like Distributed Teaching Collaboratives that broaden access to robotics education,” said Eric Krotkov, advisor, University Research Program, Toyota Research Institute. “TRI looks forward to continuing the work with Professor Jenkins.”
In 2022, the program expanded with support from Amazon and a multi-university award from the Sloan Foundation, funding a formal collaboration led by Jenkins, including Joseph, Grizzle, and Leia Stirling, an associate professor of robotics at the University of Michigan.
“We envision a not-too-distant future where it’s common for robots to be invaluable home assistants and trusted companions. As we invent on behalf of all of our customers, we seek to hire and retain a diverse, talented workforce, and recognize our responsibility to help create it,” said Darnell Moore, Ph.D., principal technical program manager for Amazon Consumer Robotics University Partnerships.
“DTC offers a compelling and scalable blueprint for building an inclusive pipeline of roboticists while also addressing some of the barriers preventing broader participation in robotics education. We’re excited to work with thought-leaders like Professor Jenkins and other faculty at Michigan Robotics to create more pathways for students and MSIs.”
Darnell Moore, Ph.D., principal technical program manager for Amazon Consumer Robotics University Partnerships.
With these resources, the team added new courses such as Applied Optimal Control with Florida A&M and Deep Learning for Robotic Perception with the University of Minnesota-Twin Cities and the Colorado School of Mines. Some courses are now run independently at MSIs, such as Computational Linear Algebra at Morehouse.
For each distributed course, the teaching faculty provide materials online so that anyone can use their work as a springboard to develop robotics classes. Each course handles its own web presence—for instance Robotics 101 is on GitHub and Robotics 102 has a website.
Ripple effects of Distributed Teaching Collaboratives
The program has also provided an avenue toward robotics concentrations or minors at partner institutions.
“There’s definitely a lot of interest in creating a program around robotics,” Perry Sweeper, a professor of practice at Morehouse, said during the first gathering of distributed teaching collaboratives faculty and student instructors last summer. “We’re looking at first steps to do that, bringing these classes and getting the information. And being here at Michigan has helped us gain a lot of that.”
At that gathering, enabled by support from Amazon, instructors sharpened their skills in the programming languages they teach. They also got on the same page about terminology for robotics concepts and tasks. This is particularly useful for students when they face homework assignments that are common across the participating colleges and universities.
Jenkins is pleased with the progress the team has made: They’ve demonstrated the value of the model. And he’s looking forward to continuing to sustain and grow it, with help from supporters. Additional instructors and permanent staff are needed to manage course resources and coordination.
“We can grow a diverse talent pool, not just at Michigan but across all our partner institutions. Yet we won’t be able to continue sustaining and expanding courses without supporters,” said Jenkins. “If industry joins us in the classroom, they can engage with a rising cohort of engineers who are ready to advance their robotics and AI technologies.”
Going into the fifth year, Jenkins sees an opportunity to engage more undergraduate students in teaching, at partner institutions as well as at U-M, particularly as Morehouse students served as TAs for the first Robotics 101 offering at Howard.
Jenkins also notes a role for collaborative teaching in related efforts at MSIs, such as building up research partnerships.
“If you’re trying to build a respectful relationship with HBCUs, the place to start is the classroom,” said Jenkins. “The classroom is the catalyst.”