|
In this paper, we have presented methods that significantly improve the robot awareness of its terrain traversability conditions.
The terrain traversability awareness is achieved by association of terrain image appearances from different poses and fusion of
extracted information from multimodality imaging and range sensor data for localization and clustering environment landmarks.
Initially, we describe methods for extraction of salient features of the terrain for the purpose of landmarks registration from two
or more images taken from different via points along the trajectory path of the robot. The method of image registration is applied
as a means of overlaying (two or more) of the same terrain scene at different viewpoints. The registration geometrically aligns
salient landmarks of two images (the reference and sensed images). A Similarity matching techniques is proposed for matching
the terrain salient landmarks. Secondly, we present three terrain classifier models based on rule-based, supervised neural
network, and fuzzy logic for classification of terrain condition under uncertainty and mapping the robot's terrain perception to
apt traversability measures. This paper addresses the technical challenges and navigational skill requirements of mobile robots
for traversability path planning in natural terrain environments similar to Mars surface terrains. We have described different
methods for detection of salient terrain features based on imaging texture analysis techniques. We have also presented three
competing techniques for terrain traversability assessment of mobile robots navigating in unstructured natural terrain
environments. These three techniques include: a rule-based terrain classifier, a neural network-based terrain classifier, and a
fuzzy-logic terrain classifier. Each proposed terrain classifier divides a region of natural terrain into finite sub-terrain regions
and classifies terrain condition exclusively within each sub-terrain region based on terrain spatial and textural cues.
|
