Terrain Traversability Estimation Of The Outdoor Moboile Robot

As outdoor mobile robots are often need to work in the unstructuredenvironment, which can be unknown or not be entirely known. However, most ofthese circumstances might have a great bad impact on the capability of the robotmotion and traverse, and then may cause the navigation task failure or evendamage to the robot itself. Therefore, researches on the terrain traversabilityhave become one of the core problems of autonomous navigation for mobilerobots. Considering the robot is certain, the goal of this research is to propose annew evaluation algorithm based on Bayesian Classifier Theory to calculatetraversability index according to the terrain feature, so that it can assist the pathplanning and provide the correct strategy to be sure the robot can be able tocomplete the autonomous navigation tasks safely and effectively.In this dissertation, it firstly proposed to extract the geometric properties(slope) and non-geometric properties (slip characteristic) common as evaluationfactors for evaluating the traversability of the outdoor mobile robot. At the sametime, some experiments were carried out to verify effectiveness of the algorithmsfor terrain characters extracting. Secondly, the power cost though the samedistance were calculated as the evaluation criteria, and based on the BayesianClassifier Theory complete the estimation algorithm of the traversability design.In order to verify the available and effectiveness of the whole algorithm, itneeds to design and build platform for experiment to obtain some experimentaldata. Through the analysis of the experimental results, the algorithm designed inthis dissertation can be used to calculate the traversability by extracting theterrain features, and get the target for the path planning and the autonomousnavigation being improved.