By established principles of mathematical physics, the velocity of an object is defined as the rate of change of its position, and the acceleration is defined as the rate of change of the velocity. These relations can be applied to the navigational problem of position finding if an instrument can be devised to measure acceleration and then to convert it successively to velocity and to position. In the terminology of calculus, acceleration is integrated (summed a little at a time) to get velocity, then velocity is integrated to get position.
In one form of accelerometer, a reference mass is suspended on springs within a housing firmly attached to the craft. The inertia of the mass causes it to tend to remain stationary, but any acceleration of the craft tends to displace the housing relative to the mass. The forces required to nullify relative motion of the mass and the housing—in three directions fixed by gyroscopes—can be measured electrically. The electrical signals are directly related to the forces and, by Newton’s second law of motion, to the accelerations. Standard electronic circuitry performs the necessary integrations of the accelerations to provide the distances and directions, in three dimensions, through which the craft has moved from its original position.
Such combinations of accelerometers coupled with integrators are called inertial guidance systems; in the context of navigation, they amount to sophisticated dead-reckoning devices. Since their introduction, starting in 1950, they have proved extremely valuable in controlling trajectories of submarines, booster rockets, and spacecraft. Their errors, like those of any other dead-reckoning system, are cumulative with time, but nuclear-powered submarines have traveled under the north polar ice cap, guided solely by inertial systems, with errors of less than a mile per week.
Dead reckoning by computer
In modern inertial navigation systems, computers have proved well-suited to processing the streams of data—directions, speeds, and times—involved in keeping track of position. In military land vehicles, computers are fed by compasses and wheel-mounted sensors. Navigators aboard ships depend on the gyrocompass and the log; those in aircraft rely on the gyromagnetic compass and Doppler-effect speed measurements. The computers can be programmed to display or print periodically updated positional information. Inertial guidance systems may provide dead-reckoning information only, though compass and Doppler data can be combined with inertial outputs. Information from radio navigation systems, such as loran or the global positioning system (GPS), can be added to the dead reckoning.
Radio navigation systems that can provide continuous indication of position are eliminating the distinction between position fixing and dead reckoning. Navigation accuracies are improving by supplying both the classical dead-reckoning data (speed, direction, altitude rates, and angles) and the continuously updated position to a computer, which determines the speed, heading, and rate of climb or descent that must be maintained to execute the flight plan. Many computers apply the technique called Kalman filtering, which weights each of the several supplied data according to its expected quality and uses previous position and velocity solutions in determining the current best estimate of position and other desired quantities.
Originally, analog computers were used in navigation systems, and calculations of a relatively simple nature involving inputs from various electrical sources were continuously performed. Today, digital computers are employed for performing virtually all the necessary calculations. The digital computer works so fast that for navigational purposes it can be considered virtually instantaneous and can therefore provide continuous information for control purposes. It has a memory to store information for use when needed. It is built from electronic modules that are mass-produced at low cost. It has only one disadvantage. Conversion of analog information into digital form can be costly. Hence, although far superior to the analog computer, it is less economical whenever a large number of electrical signals must be combined in relatively simple ways without any need for memory. Such situations still apply to control systems in many craft.