The New-World Explorer
Srikanth Schelbert (MSR ‘24) turned his final project into a toolkit to help roboticists more quickly create autonomous robots to roam new frontiers. Now he's applying lessons learned from that project to his new job at the Software Engineering Institute of Carnegie Mellon University.
Think of Srikanth Schelbert (MSR ‘24) as a modern-day Meriwether Lewis or William Clark.
Those two famous explorers led a party that ventured out to map previously unknown territory in the western United States in the early 1800s. No outsiders had been where they went, and their explorations and maps made it easier for future adventurers and settlers who followed.
More than 200 years later, Schelbert used his final project in Northwestern Engineering's Master of Science in Robotics (MSR) program to make robot exploration easier for fellow roboticists.
Schelbert’s project led to what he calls a “Frontier Toolbox,” a ready-to-use downloadable kit of exploration algorithms that can be applied to a wide array of robots that map out new territory.
“If a robot has an idea of where it is in the world, there are going to be edges in those worlds,” Schelbert said. “The idea of frontier exploration is to expand the known area of a map by navigating to the edges and pushing those boundaries.”
The challenge for roboticists is that there are many different ways to accomplish this sort of frontier mapping, each of which ends up relying on laborious and time-consuming algorithm production to turn ideas into reality. Schelbert’s final project created a package that anyone could download as a prototype to get their ideas rolling toward reality much more quickly.
The practical applications for these robots vary. While they are not exploring uncharted wilderness in the traditional sense, they very much are mapping territory they need to be familiar with to do the jobs they are created for.
For example, a robot could be placed in a sewer system where it needs to quickly and efficiently map the corridors and take note of any damage or anomalies noted during the mapping process.
“There are so many use cases where robots can just be plopped in an unknown environment and then they have to figure out what is the lay of the land,” he said. “Exploration could be part of it, but there could also be completing a task within this completely unknown environment.”
Schelbert’s final project created a variety of base algorithms that will help others get their robots up to speed faster. He did this by doing the hard and laborious work himself, having a robot traverse uncharted territory with a variety of approaches – from finding the newest unexplored frontier closest to its current position to choosing a route that would add the most new information to its mapped knowledge. Each approach employs a different algorithm.
These missions are easy for a robot to accomplish in an empty building. But just like the wilderness Lewis and Clark explored was filled with harsh terrain, wildlife, and native human inhabitants, useful robots rarely exist in empty spaces.
“A hallway is pretty easy for a robot to just go forward until it can't because there's not a whole lot to see,” Schelbert said. “But if it's in a lab space where it's cluttered and there are a lot of places where it can see stuff but it can't actually move through, that presents a challenge.”
Thanks to Schelbert’s work, it’s a challenge future roboticists will find easier to solve. Schelbert deflects credit for the project to the MSR program that helped teach him the knowledge to make his final project successful — particularly MSR co-director Matthew Elwin and PhD candidate Max Muchen Sun.
Schelbert said the MSR program’s structure gives students a common foundation at its outset and then allows them to pursue their own passions as they work toward their ultimate destination – graduation and a successful career.
Schelbert compared the MSR program to a choose your own adventure book.
“For me, it was mobile robotics, but for other people it could be underwater robotics, deep machine learning, deep artificial intelligence, or the intersection of any of these things,” he said. “What I've really benefited from the most is that flexibility to pursue my passion and create robots that can move around and complete various tasks.”
Schelbert also found a successful start to his post-MSR career. He recently started a job in the Autonomy Lab at the Software Engineering Institute of Carnegie Mellon University. Schelbert interned in the lab during his time in MSR and credited the program with helping put him and his fellow students in the best position to succeed professionally — no matter what robotics frontier they choose to explore.
“The program does a really good job of preparing you with the mindset and skill set to know how to interact with recruiters and how to interact with companies of all sizes,” he said. “If you are willing to put in the work, it’s all within reach.”
