Multi-Vehicle and Multi-Sensory Dataset with 3D Scene-Dependent GNSS Multipath

This dataset comprises the supplementary material of the paper GNSS Simulation for Automotive: Introducing 3D Scene-Dependent Multipath With CARLA (https://doi.org/10.1109/ACCESS.2025.3543752), which describes the simulation pipeline in order to obtain vehicles' sensor data as well as the raw GNSS measurements considering dynamic multipath. 

The CARLA simulator was used to generate vehicles' trajectories, with Town03 scenario. 5 vehicles equipped with GNSS, IMU, and odometer sensors with a random motion were considered. Sensor sample rate is 20 Hz and simulation duration is 1200 s (20 minutes). 

To generate GNSS raw measurements with scene-dependent multipath, satellites' orbits were considered during the CARLA simulation to detect when the satellite signals are obstructued by the obstacles in the city. This allowed us to estimate the multipath in urban areas, thus providing GNSS pseudoranges considering the 3D environment of the simulated environment in CARLA. GNSS raw measurements were generated with GPSoft SatNav Toolbox for Matlab, considering the multipath estimate from the proposed approach.

Full simulation configurations as well as the description of generated files are available in the readme.pdf file. 

Citation Request

If you would like to use the dataset provided please cite the two following items::

C. Pendão, I. Silva, F. Botelho, and H. Silva, ‘‘GNSS Simulation for Automotive: Introducing 3D Scene-Dependent Multipath With CARLA’’,  in IEEE Access, 2025, doi: 10.1109/ACCESS.2025.3543752.

C. Pendão, I. Silva, F. Botelho, and H. Silva, ‘‘Multi-Vehicle and Multi-Sensory Dataset with 3D Scene-Dependent GNSS Multipath,’’, Zenodo, doi: 10.5281/zenodo.13145215.