Home

10%

PEDAL TO THE MEDAL

The WVU RoboRacer team builds race cars that have minds of their own

By West Virginia University

March 16^th, 2026

It’s lights out, and away we go.

The smooth, curving track.

The car, crouched low above the ground, hugging the course with wide tires.

The snarling whine as it flashes past, whipping through turns.

It all comes down to this — exceptional engineering, pinpoint accuracy at high speed, and a driver who doesn’t know the meaning of fear.

That’s RoboRacer at West Virginia University.

George Harmon controls a small, yellow robot car on a concrete floor surrounded by orange tubes. He stands in a lab, looking toward the robot with a controller in his hands. — WVU senior George Harmon, a member of the WVU RoboRacer team, observes the performance of the algorithm he developed to control a self-driving scale-model race car on a track at the WVU Statler College, March 2026.

IN THE FAST LANE

RoboRacer autonomous racing is an international competition between self-driving vehicles. Students at WVU and other colleges and universities around the world develop self-guided systems for scale-model race cars, then come together to compete in an annual global throwdown.

In 2026, the RoboRacer competition happens at the Institute for Electrical and Electronics Engineers International Conference on Robotics and Automation, in Vienna, Austria, and the team from the WVU Benjamin M. Statler College of Engineering and Mineral Resources is ready.

Amr El-Wakeel, assistant professor and coordinator for robotics in the Lane Department of Computer Science and Electrical Engineering, founded the University’s RoboRacer chapter in 2024 with Computer Engineering doctoral student Rukesh Prajapati and Electrical Engineering doctoral student Mohamed Elgouhary. At the team’s first competition, they placed second.

Now, the WVU RoboRacer team has grown to include over 30 members — first-year students, engineering majors taking El-Wakeel’s Intelligent Safe Robotics course, and seniors who participate in RoboRacer as part of their senior capstone studies, along with graduate students.

Computer Science master’s student Andrew Sarver has taken a lead role in preparing the team for this summer’s race, while Elgouhary —who in his own doctoral research pushes the boundaries of autonomous systems — is a mentor to the undergraduate members.

The team works together on troubleshooting, but every member is responsible for independently developing an autonomous driving stack — the brain of the car — that they each feel is “their baby,” El-Wakeel said.

“It’s a battle of algorithms. Competitors try to develop an algorithm that can outperform others in how it coordinates perception, how it plans its path along the race line, and what control aspects you’ve given it, such as whether it can adjust its steering angle or speed. Then we put our algorithms up against each other and race.”

WVU F1Tenth Roboracer team members work in a lab, gathered around a computer and small robot car. One man types, others watch, in a brightly lit room. — The WVU RoboRacer team clusters around a workstation as they fix a problem with the algorithm that powers their self-driving race car. From left: Aaron Pander, Stephen Heflin, Amr El-Wakeel, Mohamed Elgouhary, James Kmetz, Blake Buskirk, George Harmon.

FULL THROTTLE

A RoboRacer algorithm with the potential to win it all will be able to make mid-race decisions —increasingly AI-powered — about perception, control and planning.

How that algorithm makes those decisions is up to each developer. Computer Science student George Harmon, from Charleston, West Virginia, said most team members work with “reactive” algorithms that respond to the environment but don’t store data, “memory-based” algorithms that use past experiences to make decisions, or algorithms that use a combination of those two approaches.

“One of our reactive algorithms is called Gap Follow,” Harmon explained. “In Gap Follow, the car takes in a range of scans from the sensor. It can look at those scans and say, ‘Oh, there’s a huge gap here that I could go through and not hit anything.’ Gap Follow just follows that gap.”

Once memory enters the equation, things get more complicated. Harmon also uses a memory-based algorithm called Pure Pursuit that generates its own localized map of the track as the car drives around it. The car then follows waypoints within the map, Harmon said, rather than relying on sensor scans of the environment.

Memory-based algorithms like Pure Pursuit can’t react to real-time changes in the track, like a competitor’s flipped car, but they do allow the developers to do things like tune the ‘look-ahead distance,’ based on the track geometry,” Harmon said.

“With a higher look-ahead distance, the car is looking way off in the distance. That means that in a straightaway, it can go much faster with less oscillation. When we’re going around a curve, however, we want the look-ahead distance to be a lot shorter.”

The software is the focus, but the hardware is an important partner. Recently, for example, Pander’s Pure Pursuit code was working fine, following the course of the racetrack, “when suddenly it just crashed,” he said. “But the data was right.”

Eventually, Pander realized the problem was minor.

“It was just how I set the angle of the sensor. Once I tweaked that, it was all good. The problem is finding where the problem is,” he said.

NEED FOR SPEED

The key to winning in autonomous racing, as in life, is “striking a balance between competitiveness and safety,” El-Wakeel said.

He explained that when a new student joins the RoboRacer team and begins working on their algorithm, they usually go for safety first, focusing on ensuring the car is stable.

“Then they complete their very first lap around the track at a slow speed and wow, you can see the excitement in their eyes. They start to say, ‘Okay, now let’s push the limits.’

“And then they crash. They become super cautious. They slow down when they see an opponent. They don’t try to overtake them. But bit by bit, they start to push the boundaries again.”

In addition to designing the software that runs the race cars, RoboRacer students play with the design of the track itself.

“During the development and testing stage, we can shape the layout of our track however we want,” El-Wakeel said. “We can make it to be basic and straight, or we can make a bunch of curves in it. We usually throw a couple boxes in the middle to see if we’re able to dodge those obstacles.”

When it comes to building the cars, participants work from a universal kit. Every RoboRacer car is equipped with a two-dimensional laser sensor that reads raw data from the environment.

The sensor takes in arrays of values representing the racetrack and opponents. Then the car picks the best route and goes.

Teams can make some hacks to the car’s hardware. For instance, WVU has replaced the bouncy spring suspension that comes with the kit with solid 3D-printed cylindrical pieces, designed in-house, to minimize vibration when taking sharp curves at high speed. The new suspension helps stabilize the car, as do the air vents the team introduced into the “detection box,” which is mounted on the car’s back end and allows opponents to recognize each other on the track.

WVU F1Tenth Roboracer team members work in a lab. Students monitor laptops as a small robot navigates an orange tube track. A model airplane hangs above. — On the floor of a lab at the WVU Statler College, orange boundaries mark the outline of a racing track laid out by the WVU RoboRacer team to test the software that powers their autonomous scale-model race car, which is preparing to take a hard turn.

The WVU F1Tenth Roboracer sits on a concrete track in a lab. The autonomous car has a yellow and black rear wing, visible wires, and sensors. Orange tubes line the background. — One of the scale-model race cars used by the WVU RoboRacer team revs for its close-up on the floor of a lab at the WVU Statler College.

HAMMER TIME

Ultimately, though, RoboRacer “is not a race based on hardware, but a race based on software.”

That’s according to team member Aaron Pander, a double major in Computer Engineering and Electrical Engineering from Grafton, whose experience with RoboRacer marks his first foray into robotics.

“It’s very different from what I’ve been used to,” Pander acknowledged. “I had no experience with the Linux platform we use or coding in Python, so I spent the first semester just getting familiar with the car and getting it to work. In the second semester, everyone on the team has a basic understanding of the concepts involved, so that’s when we do serious testing.”

For his teammate James Kmetz, from Silver Spring, Maryland, RoboRacer was also a chance to try something new.

“I study computer engineering, which is about the hardware of computers,” Kmetz explained. “Computer engineers like me don’t usually touch software, but I got into RoboRacer because the car has really cool and specific hardware, and because I’m a big fan of F1 racing. When I heard about this group for autonomous racing, I said, ‘Oh, yeah.’”

An aficionado of racing video games, Kmetz finds that working on his RoboRacer algorithm is a lot like playing Forza Horizon, one of his favorite game series.

“Forza Horizon has an algorithm already in place in their system that shows a player the optimal race line, the best path for their car to follow — which is just like the algorithms we’re doing in RoboRacer. The race line dictates when the car will break, when it will hug the corner. And then, if we want to factor in other cars on the track, we have to set it so that, every once in a while, if it notices something there, a car or another obstacle, it will react to it and veer off the race line slightly before it returns to trying to follow it again.”

James Kmetz of WVU F1Tenth Roboracer team smiles, holding a small robot car. The lab setting includes orange tubes on the floor, and a second robot car. — With the WVU RoboRacer track behind him, Computer Engineering major James Kmetz holds up the RoboRacer team’s scale-model, self-driving race car.

GREASED LIGHTNING

When the rubber hits the road, the cars often switch back and forth between reactive and memory-based algorithms, and sometimes rely on programming that incorporates elements of both approaches.

Harmon, for instance, is working on “model predictive control, which is really cool because it creates a model of the car. I have to figure out how fast can the car go before it drifts, what’s the lateral acceleration limit before the car slips — just a lot of experiments that I have to run to get a good model of the car,” he said.

“But once I have that model, the model predictive control algorithm will essentially predict all future possible trajectories, and it will choose the best one that doesn’t collide with anything and follows the race line.”

While Harmon fine-tunes his algorithm by running simulations of situations it might encounter on the racetrack, his teammates Blake Buskirk, from Honolulu, Hawaii, and Stephen Heflin, from Clarksburg, emphasized that simulations rarely replicate all the ways the real, three-dimensional world affects a car’s performance.

Heflin, for instance, worked on an algorithm called Path Follow, based on “disparity lines” — distances the car can’t see past, which Path Follow attempts to extend geometrically. When he moved testing from a simulated digital environment to the physical hardware of the car, Heflin said the vehicle immediately began oversteering.

Buskirk explained that’s because simulations don’t precisely model the advanced physics of the real world — the torque on the car’s wheels when they’re rotating, the effect of drag on steering, the resistance on the actual car.

“Without those physical effects, I can run my algorithm in the simulation and watch my car go straight and then take the curve,” Buskirk said. “But if I were to do it on the real track at a real speed of, say, eight meters per second, and it turned like it did in the simulation, the tires would come up because of the forces moving them, and the car would flip.”

Harmon laughed and sympathized with his teammates.

“There have been times where I’ll be looking at an algorithm in the simulator, and we’ll be going nine, 10 meters per second, and it’s great,” he said. “Then I run it on the car, and as soon as I go — it just crashes straight into the wall.”

Aaron Pander of the WVU F1Tenth Roboracer team stands in a lab controlling a small, yellow and black race car on a course marked by orange tubes. — Aaron Pander, a double major in Computer Engineering and Electrical Engineering, had never experienced robotics before joining the RoboRacer team. Here, he watches his algorithm guide a self-driving race car around a track at the WVU Statler College.

Close-up of computer screens displaying code and a track layout, likely for WVU F1Tenth Roboracer team. The lower screen shows lines of code and data tables, while the upper screen displays a white track outline on a dark background. — RoboRacer is a race based on software, not hardware. These computer monitors offer a peek at the code behind the autonomous navigation, as well as a two-dimensional view of the track as captured by the race car’s laser sensor.

FAST AND FURIOUS

By competition time, little wrinkles like crashing into walls will have been ironed out. And as a seasoned RoboRacer competitor, WVU will enjoy certain advantages, according to El-Wakeel.

“We have watched other teams crash and break down, and we have faced challenges ourselves,” he said.

At the 2025 IEEE Conference competition in Atlanta, for example, the crowd of more than 10,000 people overwhelmed the internet connection, making it difficult for the WVU team to stay in touch with its vehicle. Now, the car is equipped with highly stable antennae, ensuring that problem won’t happen again.

Similarly, after witnessing a few competitors wipe out, the Mountaineer vehicle is now designed to minimize damage in a crash, with critical components like power boards and sensors tucked away in protected locations less vulnerable to impact.

When WVU heads to the next competition, the team will first have to complete the qualification phase, in which each vehicle completes timed laps of the track on its own, judged on the number of laps it can complete without crashing and the fastest lap time.

“The main challenge there is the geometry of the racetrack itself,” El-Wakeel said. “It’s about the curves that the vehicles navigate, and the stretches where they can overtake opponents in the next stage. That tests the stability and adaptability of the cars and their algorithms.”

Then comes the head-to-head competition, in which two teams at a time race against each other, hoping to avoid crashing into the racetrack or each other. If a crash does happen, a member of the team must leap onto the track, avoiding any interaction with the other cars, retrieve the vehicle, and attempt to repair it and get it back in the race.

With hard work and a little luck, El-Wakeel believes WVU can hold its own among the other competitors.

But for him, the race isn’t the point.

“It’s not about the competition,” he said. “This really is about finding fun and excitement in your research and experiential learning while collaborating with a team. We are developing advanced hands-on software and robotics skills in a rising generation of engineers and computer scientists who are going to make our cars and roads safer.”

APPLY NOW REQUEST INFO VISIT WVU

Footnotes

Story by Micaela Morrissette. Photos and video by Scott Lituchy.

FootNotes

Story by Micaela Morrissette. Photos and video by Scott Lituchy.

© 2026 West Virginia University

A WVU neuroscience student finds her head space

An Accidental Journey

For a WVU education alum and children’s book author, whimsy nurtures learning

MIDDLE SCHOOLERS PUT PRESIDENT BENSON TO THE TEST

March 5^th, 2026

The Experimental Tourist

WVU roboticist Yu Gu lets his curiosity lead him — all the way to outer space

FINDING THE RIGHT KEY

Mon Hills Records students raise their voices

MARRIED TO THEIR WORK

Home and professional lives intersect for WVU faculty couples

Loading, please hold on.