Racing Line Optimization
As part of the Formula Student team at IIT Madras, we were developing a laptime simulation program to be used to evaluate different design concepts, components and most importantly be a central tool for designing the first electric powertrain by the team.
There are two parts to the laptime simulation program, the car model and the track model. I focused on extracting the track boundaries from birds eye views of tracks and successfully obtained the racing line for multiple tracks.
Laptime Simulation Flow
Starting from an overhead schematic/picture of the track, the track boundaries are extracted. These boundaries are then fed into a Racing Line Optimization program which calculates the path of least time that the car takes. Now different car models (design concepts) can be tested on the track.
The Racing Line Optimization program moves around a fixed number of waypoints that are interpolated to give the full racing line. The Global Optimization toolbox in MATLAB provided the solver for this optimization with the objective function set to be the laptime (the quantity to be minimized).
Surrogate optimization creates a surrogate model of the computationally expensive objective function and minimizes the surrogate model. The surrogate model is obtained through random sampling of the objective function. This process is repeated (unless something goes wrong!).
A sample velocity map of the Formula Bharat 2020 track organized at the Kari Motor Speedway in Coimbatore, India in January of 2020. This is the result after extracting the track from an image, and running the electric car concept as a single track bicycle model on the racing line obtained from the racing line program.
All figures are mine.
Picture of the FSG track courtesy of FSG Media - found in the FSG2019 magazine.
