TUfast at the Formula Student Electric
The Formula SAE was founded in 1979 by professors in the United States, and came to Europe in 1999. The project aims to let students challenge themselves, test their abilities, and learn how to work on a huge project in a team. The competition has welcome the Electric category a few years ago and TUfast competes with its electric car named “eb018” which embeds the Ellipse2-N, a miniature Inertial Navigation System from SBG Systems.
The Ellipse2-N INS/GNSS has been installed on the eb018. The IMU and GPS speed are the main sources of the filter the team used to estimate the state of the vehicle (Speed, Slip angle, X and Y accelerations and Yaw Rate). This state has been then compared to a desired state to generate the command of each motor. The Ellipse2-N was therefore a very decisive factor in the TUfast very successful 2018 car (1st place at autocross in UK, Germany and Spain; 1st place overall in Australia).
The GPS positions were extensively used for analysis. The team generated many maps to understand more intuitively all phenomenon influencing eb018’s performance. A very insightful example is the track map below. It shows an internal correction factor in our Kalman filter. We learn with it that our tire model overestimates longitudinal tire forces (blue/green in straights) and has a rather good estimation of the lateral forces (orange/yellow in corners).
Ellipse2-N Use in the 2019 Car
The Ellipse and its software were great to work with and easy to configure. The documentation contains everything we needed to get started and develop the interface with our system. The data provided by the Ellipse is accurate and has shown at some moment less than 10cm of error on 1km+ trips. We are very satisfied with this product explains Alexandre Kopp in charge of Vehicle Dynamics at TUfast Team.
On eb019, we will exploit the potential of the Ellipse2-N even more. We will use 2 Kalman filters; one for the main state (as on eb018); and a second one to filter the sensors and data fed to the physical model of the main filter. The main filter will also be improved with a position and heading estimation. The Ellipse2-N thus remains the most important sensor of our car for the state estimation. Its integrated Kalman filter will be especially useful on eb019 with accurate estimations of the roll, pitch and yaw angles. These three are necessary for the aero forces’ calculations.