top of page
_JDT2705.jpg

About the project

I participated on the development of the first autonomous vehicle in the history of UPC ecoRacing. My contribution was on the development of the perception system, more specifically I designed and developed the HW and SW from the perception system.

The perception pipeline consisted in single sensor algorithms with the camera, single sensor algorithms designed for the LiDAR, and a combination of both sensors, making an algorithm for the sensor fusion.

The first one of the previous algorithm consisted on the detection of the cone's bounding box using YOLOv3 and using the extrinsic calibration w.r.t. LiDAR the base of the bounding box (in contact with the ground) is used to approximate the cone's distance, using the homogrphy projection matrix.

The second algorithm, based on LiDAR, was based on pointloud manipulation and clustering, loosing the color information of the cones.

Finally the fusion of both combined the good distance detection of the LiDAR and the good colur detection of the cameras. It consisted on the projection from the pointcloud in the image and discriminate the cluster position based on the bounding box.

Oficial ecoRacing's webpage

  • 5H5DPEhJ_400x400

View Gallery

Victor Escribano Garcia's Portfolio

  • alt.text.label.LinkedIn
  • GitHub

©2024 by Victor Escribano Garcia's Portfolio. Proudly created with Wix.com

bottom of page