Contents I. INTRODUCTION
The variation of outdoor environment such as weather, illumination, and changes of vegetation characteristic cause difficulty in developing a navigation system for outdoor autonomous mobile robot that is able to navigate all the time in all conditions. Even if several researches have been done in this field so far, it still remains as a major challenge. Generally, GPS is well known as one of the most feasible position measuring sensor for outdoor navigation. However, GPS does not currently offer a sufficient navigation resolution to allow it to be used as a standalone global navigation method for mobile robot. It is because of the ionosphere and the troposphere which may cause the error of time delay of signal from satellite and multi-path phenomenon which occurs when the GPS signal is reflected off objects such as tall buildings or large surfaces before it reaches the receiver. Thereby, this only allows GPS to act as a coarse position fixing aid to other localization methods to get an adequate position. Ohno et al. [1] present a differential GPS and odometry-based outdoor navigation of a mobile robot. The mobile robot actually localizes based on odometry and the positions obtained from differential GPS are used to correct the cumulative error made by odometry. Odometry is also used to discard the erroneous differential GPS data. However, failure in discarding erroneous differential GPS data and interception of GPS signal cause failure of navigation. On the extended work, Moracles et al. [2] present autonomous robot navigation in outdoor cluttered pedestrian walkways. A landmark recognizing method is used to solve problems in [1] where the landmarks such as trees and buildings are scanned by laser scanner sensors. Howard et al. [3], Burguera et al. [4] and Guivant et al. [5] also present the similar method of landmark recognizing using laser scanner sensors. Since such landmarks are not unchanged and on the other hand, in the pedestrian walkways the obstacles such as people, bicycles and cars may appear during the navigation, mismatching of landmarks can be occurred. Botenstein et al. [6] provide an idea of using magnetic compass to compensate the localization based on landmarks recognizing and Chiaju et al. [7] provide a method of using magnetic compass to compensate the localization based on GPS to make the navigation system be more reliable. However, magnetic compass is distorted near power lines or steel structures [8]. Odometry-based navigation can also be compensated by gyroscope to approve its accuracy. For instance, Brenstein et al. [9], [10] present a new method for combining data from gyros and odometry and an experimental evaluation of a fiber optics gyroscope for improving dead-reckoning accuracy in mobile robots. The method in [9] and [10] works fine in the indoor environment with a reliable accuracy. However, for outdoor environment where the terrain is uneven the cumulative error of orientation of mobile robot can be occurred. The vision-based navigation is also well known as one of the fundamental skills for outdoor environment. Lynch [11] provides experimental evidence of the extent to which people use landmarks when finding their way around cities. Morita et al. [12] present panoramic view-based navigation in outdoor environments based on support vector learning. The robot is placed in positions to be recognized where the image features such as color and edge are extracted and then sent to a set of SVMs, each of which is trained to recognize objects of a specific class. The output vectors from all SVMs are concatenated to produce the final recognition result. The interesting point is only the upper-half part of the image where the unmoved objects such as trees or buildings exist is used in the method. Konolige et al. [13] provide outdoor mapping and navigation using stereo vision. A 3D map is built relied on visual odometry to provide good localization, ground-plane analysis to help detect obstacles and sight lines to identify distant regions that are likely to be navigable. The robot is possibly able to locate a goal position several hundred meters away. However, water and ditches are two robot-killers and augment from GPS for global consistency is required, because it would give finer adjustment in the robot position and safer navigation.
