Tzu-Yi Chiu, Narimane Zennaki (Polytechnique Montréal)
The script to be executed is main.m with Matlab/Octave (both compatible).
One can manually set the boolean variables do_plot, do_plot_ellipse
(covariance) and do_plot_lidar (lidar detections matched with landmarks, and
initialization of landmarks) for different levels of visualization, at the cost
of the running speed.
The figure shows the evolution of the vehicle position and the map of trees
being constructed, while also showing the received GPS signals.
do_cal is set to true if one wants to obtain the error evolution, comparing
the vehicle position to the GPS data.
-
aa3_dr.mat,aa3_gpsx.matandaa3_lsr2.matare the Victoria Park datasets, corresponding to odometry, GPS and LIDAR respectively. -
info.txtcontains the format information relative to the dataset. -
main.mis the main script to be executed. -
detectTreesI16.mis provided along with the Victoria Park dataset, which allows to transform LIDAR raw data to measurements.detect.ma post-processing function for our own purpose. -
VictoriaEKF.mis a class which implements the EKF. -
munkres.mis a function which implements the Hungarian algorithm, allowing to compute data association. Borrowed from: https://www.mathworks.com/matlabcentral/fileexchange/20328-munkres-assignment-algorithm
The following figure shows the constructed map, where green crosses indicate trees, while red points indicate received GPS signals. The black line shows the estimated trajectory, and the current estimated position is shown with a black circle.
