Depth Camera and Laser Sensors Plausibility Evaluation for Small Size Obstacle Detection

Abstract

Detecting small-size obstacle objects in an autonomous vehicle or robot driving is considered an important issue in collision avoidance especially in applications such as robot navigation or vehicle parking. In this paper, an effort is made to evaluate the distance detection fusion of a target away from two uncorrelated, different technology sensors. The Robot Operating System framework is used to manage data collection between the two sensors and the python software engine. The negative behavior of the single-pixel distance detection of the depth camera was concluded according to the measurement tests creating a plausibility problem for the obstacle detection process as a fused sensor. A solution is also proposed in this paper to overcome this problem using center pixel averaging. The plausibility check algorithm was successfully proposed and tested practically detecting the implausible distance measurements.