Evaluation of Multiple Correction Methods for Depth Images for UAV Implementations

Abstract

Accurate perception of the local environment
is essential for autonomous navigation, and depth
information is crucial for path correction. Such information
can be obtained using various sensing devices, like
Ligth Detection and Ranging (LiDAR) sensors or, as
in this work, stereo cameras. However, raw sensor
measurements are often subject to error, which can
lead to incorrect decision-making. To address this, it
is necessary to develop error characterization techniques
and correction algorithms that enhance the reliability
of autonomous navigation systems.