Читать книгу Position, Navigation, and Timing Technologies in the 21st Century - Группа авторов - Страница 42

Electromagnetic radiation at wavelengths within the visible range, which extends approximately between 380 and 750 nm, as well as in its lower or upper vicinity, known as ultraviolet (UV) and infrared (IR) light, are part of some of the most common indoor positioning systems that use wireless technology.

Visible light systems typically utilize general‐purpose cameras and have been adopted particularly for indoor localization of robots. One common approach is to have a robot carry a camera to capture images of the environment that can then be processed to infer location with respect to the environment [7]. Other approaches deploy cameras in fixed locations across the environment, and if the salient features of the object to be tracked appear in the field of view of the camera, the location of the object can be calculated with respect to the camera's fixed position [8]. A key challenge is how location can be estimated in a 3D world when the primary observations are 2D, from an image sensor. Depth information can be obtained by making use of the motion of a camera. In such an approach – known as synthetic stereo vision – the scene is observed sequentially from different locations by the same camera (or by multiple fixed coordinated cameras) and image depths can be estimated in a manner similar to the well‐known stereo‐vision approach. Alternatively, distances can be directly measured with additional sensors, such as with laser scanners or range imaging cameras. The latter returns a distance value for every pixel of an image at a specific frame rate.

Figure 37.1 Taxonomy of signals for indoor localization.

All visible‐light‐based localization approaches require some form of image processing, which is time consuming and can be particularly error prone in some dynamic environments, for example, due to illumination variability [8]. In the case of laser based‐solutions, only class 1 laser devices should be used, which are classified as “eye‐safe” by the IEC 60825‐1 standard [9]. Another challenge arises due to occlusions caused by dynamic elements of the environment (e.g., moving objects or people). One way of reducing occlusion is to deploy sensors with overlapping coverage areas [8]. However, clinical settings and public indoor areas such as shopping malls are often densely populated, and therefore occlusion conditions can arise frequently even with ceiling‐mounted sensors.

IR‐based localization systems are also very popular, relying on a LOS communication mode between the transmitter and receiver. For instance, [10] presents an IR‐based localization system for museums with IR emitters installed in the ceiling of the door frames of every room. Each emitter transmits a unique ID using the Infrared Data Association (IrDA) protocol. Visitors carry a personal digital assistant (PDA) with an infrared port. The PDA contains a database of visual and textual information of the exhibits, as well as maps of the museum. Upon reception of a new ID, the PDA automatically presents the map of the corresponding exhibit hall. The main advantage of using IR‐based system devices is that they are small, lightweight, and easy to carry. However, IR‐based indoor positioning systems also have several limitations for location estimation, such as interference from fluorescent light, sunlight, as well as noise and reflections [11].

It is important to mention the privacy issues that may arise when using imaging‐based localization solutions. Typical solutions capture images of the environment, and thus can reveal important information about the person wearing the system or bystanders, for example, patients and health care personnel in the vicinity, in a hospital environment. This is particularly challenging because certain facilities (e.g., for health care) are required to protect the privacy of personnel, patients, and clients. The scenario becomes even more complex if image‐processing‐based mobile localization devices are designed to send captured images to central, computationally powerful servers for image processing. The confidentiality of the image is at great risk of being compromised while in transit over a network [12].