Introduction
Navigation is a key capability of mobile robots. Navigation consists of localization and path planning. The goal of localization is to determine the “pose” (position and orientation) of a mobile robot relative to its environment. There are three different localization problems in mobile robotics: position tracking, global localization, and the kidnapped robot problem. Position tracking requires the mobile robot’s starting position; it is also referred to as local localization. Global localization determines the pose of the mobile robot without any prior information. The kidnapped robot problem describes a situation in which the mobile robot is moved to another location without its knowledge. It is often used to assess how well a robot can correct localization errors. Methods for global localization of a robot can be modified so that they can also solve the kidnapped robot problem.
For mobile robots with tires, odometry sensors are typically used to solve the localization problem. Odometry sensors provide information about the robot’s movements; however, erroneous sensor readings accumulate over time. Thus, odometry is accurate enough for small-scale local movements but cannot be used for longer-term position tracking or global localization.
Additional sensors, such as a laser scanner and a camera, provide information about a mobile robot’s surroundings. Various methods have already been used in research to calculate the pose of a mobile robot. A drawback is that laser scanners are very expensive, and cameras require significant computing power for image processing. These methods also require a map and, typically, a specified starting position. If the robot’s starting position is unknown, its pose must be searched for across the entire map, which has proven difficult in large environments and takes a very long time. Global pose determination using Wi-Fi technology can support such methods by calculating the starting position. Furthermore, the “abducted robot” problem can be solved, as a localization error can be detected and a corrected pose calculated.
Today, mobile robots are often equipped with IEEE 802.11 Wi-Fi cards to communicate with computers or other mobile devices. Likewise, many buildings are equipped with Wi-Fi infrastructure, as it is a widely used and cost-effective technology. Many Wi-Fi cards can measure the signal strength of received packets as part of their standard operations. The signal strength varies noticeably between different locations in the environment. For this reason, the position of a mobile robot can be determined using signal strength. Signal strength measurement via Wi-Fi represents a cost-effective alternative to other global localization methods.
The project focuses on the localization and position tracking of mobile robots. It utilizes the Wi-Fi infrastructures available within buildings. In doing so, existing methods are expanded in two key ways. First, in addition to the robot’s position, its orientation is also determined. This is done by comparing the measured signal strengths from a robot’s omnidirectional antenna and directional antenna with those stored in a database (called a “radio map”). Second, the number of reference points in the database (Radio Map) is minimized using interpolation methods. The localization algorithms combine proprietary developments with existing methods. These methods can thus support the navigation of a mobile robot by determining a global starting position.