Considering the wide spectrum of situations that it may encounter, a robot navigating autonomously in outdoor environments needs to be endowed with several operating modes, for robustness and efficiency reasons. Indeed, the terrain it has to traverse may be composed of flat or rough areas, low cohesive soils such as sand dunes, concrete road etc... Traversing these various kinds of environment calls for different navigation and/or locomotion functionalities, especially if the robot is endowed with different locomotion abilities, such as the robots WorkPartner, Hylos, Nomad or the Marsokhod rovers. This paper presents a method for estimating on-line the navigation mode to be apply on a rover endowed with several ways of achieving navigation in an outdoor environment. It is based on a Hidden Markov Model which uses two kinds of information: context observation (classification of the terrain on the basis of the evaluation of a difficulty) and behavior information provided by monitors. Three of them have been developed and presented, many others may be added for the benefit of the whole system. The experimentation show the interest of such an approach, especially when context observation was not able to make the right assumption about the terrain, which is exhibited by the detection of a behavior issue.
Selection and monitoring of navigation modes for an autonomous rover
Auswahl und Überwachung der Navigationsmoden für ein autonomes Geländefahrzeug
2008
10 Seiten, 9 Bilder, 8 Quellen
Aufsatz (Konferenz)
Englisch
SAE Technical Papers | 2003
|Mars rover autonomous navigation
Tema Archiv | 2003
|Mars Rover Autonomous Navigation
British Library Online Contents | 2003
|Autonomous planetary rover navigation via active SLAM
Emerald Group Publishing | 2018
|