Mapping and Localization

Introduction

At least one of the primary obstacles to making intelligent robots useful and reliable is the uncertainty of robot positioning. After all, mobile robots cannot operate effectively if they do not know where they are. Accurate positioning knowledge is necessary to create high-resolution maps and accomplish repeatable, accurate path following needed for high-level deliberative behavior such as systematically searching or patrolling an area. Outdoors, use of the global positioning system (GPS) has provided mobile robots with a crutch that makes it much easier to keep track of where they are. Indoors, however, GPS is not available. To address this challenge, the INL has been working with Stanford Research Institute International (SRI) to develop new localization methods that use sampling of range readings from a scanning laser and ultrasonic sensors to reason probabilistically about where the robot is within its internal model of the world. The goal is to solve the fundamental problem of indoor positioning for a wide variety of small, unmanned ground vehicles.