Modified probabilistic laser sensor model to reduce the effect of the mixed pixel for robust autonomous mobile robot navigation

Abstract

The reliable performance of the autonomous mobile robot depends upon the accuracy and reliability of the sensors. Mixed pixel problem is the consequence of the effects caused by diverse intrinsic and extrinsic parameters, impulse noise, variation in incident angle and beam width that reduce the reliability of the laser scanner that leads to generate uncertainty in sensory information. The objective of this study is to modify the laser sensor model by reducing the effect of the mixed pixel problem corresponding to incident angles and intensity by a newly designed algorithm - Edge Detection Technique (EDT) for laser scanner to generate an efficient mobile robot mapping for the robust autonomous navigation. Various real-world experiments have been performed to check the reliability of the proposed EDT algorithm linked with probabilistic laser sensor model fitted on a mobile robot and the obtained results are validated through qualitative and quantitative analysis corresponding to conventional approaches.