model might consist of the locations (orbits) of the GPS satellites.
During the predict phase, the prediction algorithm determines what the system expects to observe based upon the world model and the current navigation state, annotated as the “Prediction Algorithm” box in Figure 35.1. During the observe phase, the system receives a noise‐corrupted measurement from the real world. During the compare phase, the predicted measurement is compared to the actual measurement. Any discrepancies are used to improve the navigation state and possibly the model of the world.
Consider a simplified example in which a user attempts to determine the distance to a wall. Perhaps the user predicts the distance to the wall is about 30 feet based upon mere eyesight to judge the distance. (The navigation state is 30 feet with much uncertainty.) Then, suppose a precise laser range finder is used to measure, or observe, the distance as 31.2 feet. Next, the prediction is compared to the observation. The user quickly dismisses the prediction and trusts the observation, because the user observation was viewed as being a more reliable estimate of distance than the prediction. Likewise, examples could be drawn which highlight the prediction heavily outweighing an observation.
Figure 35.1 General navigation framework.
The most interesting applications involve a blending of the prediction with the observation. Typical GPS applications use a Kalman filter to perform the predict–observe–compare cycle. The world model consists of GPS satellite locations. Based upon some prior information, the receiver predicts the user’s location. The observations might consist of ranges to each satellite in view. These observations are compared to a prediction of what the ranges should be based upon the receiver’s estimate of position (and assumed knowledge of the world). The system conducts a blended comparison based upon the relative quality of the predicted navigation state and the observations.
In Figure 35.1, the arrow labeled “world model updates” indicates that the world model can be changed based upon the measurements that have been taken. Some navigation systems, particularly those which are designed and deployed specifically for navigation, do not require the end user of the system to be involved in this part of the process. For example, in GPS, the world model consists of information about the satellite orbits (ephemeris), the satellite clock errors, and details that are given in the signal specification (frequency, chipping rate, etc.). The GPS system uses its own receiver network on the ground to estimate satellite orbits and clock errors and to monitor the signals coming from space, and measurements from this network are used to continually update the GPS world model. As a result, the user simply obtains the most recent ephemeris and satellite clock terms and uses them for positioning. In this way, the user is completely uninvolved in the updating of the world model, which is helpful, because it greatly reduces the complexity of the system for the user.
Unlike man‐made signals, natural signals do not generally have a dedicated part of the system that is continually updating a concise world model which describes how sensed measurements relate to the real world. As a result, the challenge with such systems is often to determine a usable world model. For example, it is very easy to obtain images of the nearby environment using a camera. However, in order to determine position and/or attitude from this kind of measurement, the user must have knowledge of what the world looks like as a function of position and attitude (the world model).
35.1.1 What Is a Navigation Sensor?
The physical sensor, depicted as the yellow block in Figure 35.1, is a critical part of any navigation system, and selection of the right sensor or combination of sensors is one of the most important decisions a navigation system designer can make. What comprises a navigation sensor?
At a basic level, any physical sensor that measures something which changes when the sensor is moved is a potential navigation sensor. Additionally, since clocks are an integral part of many navigation systems, we also consider clocks in this section as well. In contrast to a navigation sensor, which measures something that changes when the sensor is moved in some way, a clock is a sensor that measures how time “moves.” A summary of the major sensors covered in Volume 2 is given in Table 35.1.
Table 35.1 Sensors covered in Volume 2
| Sensor | Sensed phenomenon | World model required | Other considerations |
|---|---|---|---|
| Cellular RF receiver | Cellular phone RF signals | Positions of cell towers, signal timing | Example of signal of opportunity (SoOP), reference receiver sometimes required |
| Terrestrial beacon receiver | Navigation signals from terrestrial beacons | Beacon locations, signal structure, signal timing | Requires dedicated infrastructure, more design flexibility than SoOP |
| Digital TV receiver | Digital TV signals | Transmitter locations, signal timing | Example of SoOP, reference receiver sometimes required |
| Low‐frequency receiver | Low‐frequency RF signals | Transmitter location or direction of arrival, local distortion effects | Susceptible to local distortions, generally less accurate than higher frequency/wider bandwidth signals |
| Radar | RF signals | Locations of identifiable RF reflectors for absolute positioning | Generally larger/higher power than receiver‐based systems |
| Low‐Earth orbit (LEO) satellite receiver | Signals from LEO satellites | LEO satellite position/velocity, signal timing (in some cases), atmospheric models | Greater geometric/signal diversity and higher received power than GNSS |
| Inertial | Rotation and specific force | Gravitational field | Dead‐reckoning only – drift normally requires update |
| GNSS | RF signals from satellites | Satellite ephemeris and clock errors, atmospheric models | Ideal for updating inertial |
| Magnetometer | Magnetic field (including variations) | Magnetic field map | Local (vehicle) effects calibration may be required |
| LiDAR | Range and intensity of laser returns | Shape/location of objects being sensed | Can be used in dead‐reckoning or absolute modes |
