Robot Mapping and Navigation by Fusing Sensory Information

Abstract

Mobile Robots, Perception & Navigation successful mobile robot dynamic navigation in a large and complex environment because it enables the robot to function autonomously, intelligently, purposefully, and robustly. The term SLAM was first coined to describe a technique used by robots and autonomous vehicles to build up a map within unknown environment while at the same time keeping track of its current position. This technique has attracted immense attention in the mobile robotics literature and has been applied successfully by many researchers SLAM has not yet been fully perfected, but it is starting to be employed in unmanned aerial vehicles, autonomous underwater vehicles, planetary rovers, and newly emerging domestic robots. All the numerous methods proposed in literature are based on some sort of incremental integration: a newly acquired partial map is integrated with the old maps. To integrate the partial map obtained at each sensing step into the global map of the environment, the localization of the robot is fundamental. To perform localization, it needs to estimate both robot's pose and obstacles positions are needed. Map building in an unknown dynamic environment has been under study for a long time and many different approaches have been developed and evaluated (Borenstien & Koren, 1991a; Other important issues related to navigation of an autonomous mobile robot are, dealing with moving obstacles/objects, and fusing sensory information from multiple heterogeneous sensors. These issues usually cannot be resolved through the use of conventional navigation techniques. During real time simultaneous map building and localization, the robot is incrementally conducting distance measurements. At any iteration of map building the measured distance and direction traveled will have a slight inaccuracy, and then any features being added to the map will contain corresponding errors. If unchecked, these positional errors build cumulatively, grossly distorting the map and therefore affect the robot's ability to know its precise location. One of the greatest difficulties of map building arises from the nature of the inaccuracies and uncertainties in terms of noise in sensor measurements, which often lead to inaccurate maps. If the noise in different measurements is statistically independent, a robot can simply take multiple measurements to cancel out the effects of the noise. But, the measurement errors are statistically dependent due to odometry errors that accumulate over time and affect the way that future sensor measurements are interpreted. Small odometry errors can have large effects on later position estimates. There are various techniques to compensate for this, such as recognizing features that the robot has come across previously and re-skewing recent parts of the map to make sure the two instances of that feature become one. For the last decade, the field of robot mapping has been dominated by probabilistic techniques for simultaneously solving the mapping problem and the induced problem of localizing the robot relative to its growing map and accordingly different approaches have been evolved. The first category includes approaches that employ Kalman filter to estimate the map and the robot location Another approach is based on Dempster's expectation maximization algorithm This category specifically addresses the correspondence problem in mapping, which is the problem of determining whether sensor measurement recorded at different points in time correspond to the same physical entity in the real world. The Extended Kalman Filter (EKF) has been the de facto approach to the SLAM problem. However, the EKF has two serious deficiencies that prevent it from being applied to large, real-world environments: quadratic complexity and sensitivity to failures in data association. An alternative approach called Fast-SLAM is based on the Raowww.intechopen.com