Carnegie Mellon Racing · Formula SAE Electric
Driverless Controls and Path Planning
Carnegie Mellon Racing builds an electric Formula SAE car that also drives itself. I lead two pieces of it: how the car turns torque commands into deceleration, and how it decides where the track goes.
Roles
Driverless Path Planning
Delaunay Edge Selection Lead
Oct 2026 – present
Own the step that turns perceived cone positions into the centreline the controller plans against, and the robustness of the selection that produces it.
Driverless Controls
Regenerative Braking Lead
Sept 2026 – present
Own how the car converts pedal and torque commands into deceleration: the control strategy, the firmware that runs it, and the analysis that says how much braking it actually delivers.
Write-Ups
Driverless Path Planning · Oct 2026
Delaunay Edge Selection
The path planner was pairing up the wrong cones and steering the car off line. Fixing it meant instrumenting the solver, correcting three geometric assumptions, and searching over whole paths instead of one edge at a time.
- C++
- ROS 2
- Computational Geometry
- Linux
- Docker
- Simulation
Sim-validated
Driverless Controls · Oct 2026
Regen Braking Dynamics Modelling
Fitting a braking model to a season of vehicle logs to find out how much deceleration a newton-metre of regen actually buys. The answer is a range, and the reason why is the interesting part.
- Python
- NumPy / SciPy
- Robust Regression
- Vehicle Dynamics
- Data Analysis
Analysis
Driverless Controls · Sept 2026
Regenerative Braking Control Design
Two ways to give a driver regen: under the accelerator, or added to the brake pedal. I designed both, validated the first against logged vehicle data, and the team races the second.
- C
- STM32 Firmware
- Control Design
- Python
- Matplotlib
On the car