With a primary goal of increasing the points gained for the acceleration event in FSAE preliminary estimations pointed towards decreasing excess tire slip. I conducted research on the various launch and traction control methods, as well as the different ways to model or evaluate their effectiveness. Our team produced a basic python model with weight transfer to simulate an acceleration run to trial different control methods, such as a PID loop, a lookup table, and others. Modeling tire slip seems simple at first glance but many external factors introduce error. The model could be refined to consider those conditions, but real-world test data quickly and effectively allows us to compare different control methods with considerations for those errors.
Wheel slip is calculated from a high precision Vectornav IMU/GPS unit and resolver feedback. The launch control portion takes commands from the dash to stage a launch and features a simple delay or a time-based torque ramp from a lookup table. The traction control portion reduces the torque commanded from either a PID control loop or a slip % * time lookup table. Both were implemented on our main vehicle control unit and are enabled/disabled by the driver.
Although the model showed the PID control loop as having the best results, tuning the P, I, and D terms to match the real world’s conditions proved to be much more challenging and inconsistent. The simple delay launch control coupled with the slip % * time lookup table traction control showed the greatest drop of 0.1 seconds in overall acceleration time. Basic tests on a wet skid-pad proved the traction control alone to be effective, but further development is needed to refine its reactions and include other variables such as steering input and individual wheel speeds.
Early test proposal for LUT control loop only
Modeling complexity (tire grip, mechanical latency, surface and weather differences, etc.)
Test precision due to external factors
State machine implementation for on-the-fly mode switching
Simple look-up table for tuning & testing ease
~0.1s overall time decrease, 4.8s to 4.7s
Modular and adaptable architecture for future development