Occluded Volume based LIDAR-RADAR Fusion Scheme for Occluded Pedestrian Detection

We propose occluded volume based reasoning method. The proposed method handles quickly occluded pedestrian detection and occluded volume generation accurately as well as general object detection. Occluded volume means an obscured area by the obstacle. Objects in occluded volume are not detected by LIDAR. On the other hand, RADAR is possible to perceive moving object in the volume. Occluded object is estimated to pedestrian by means of unique Doppler pattern of human obtained from RADAR. In addition, the proposed method has a low processing computation in comparison with learning methods because it generates precise fusion ROI (Region of Interest) by combining ROIs of each sensors. Therefore, occluded pedestrian can be estimated to use fusion ROI and occluded volume with the characteristic of human observed by sensors.

sk

People:

  • Seong Kyung Kwon(sk_kwon@dgist.ac.kr)
  • Eugin Hyun (braham@dgist.ac.kr)
  • Jin-Hee Lee (jhlee07@dgist.ac.kr )
  • Jonghun Lee(jhlee@dgist.ac.kr)
  • Sang Hyuk Son (son@dgist.ac.kr)