Tutorial on Inverse Mechanization

Inverse mechanization converts position, velocity, and attitude (pose) data into inertial measurement unit sensor data (specific forces and rotation rates). It removes the need for expensive, real-world flights just to get reasonable sensor recordings for inertial navigation simulations. This can be helpful when real pose data is available but no inertial sensor data is included. Actually, the pose data itself could be synthetic. The researcher can then use this estimated sensor data to forward mechanize and get pose data, which should exactly match the original pose data. After generating the sensor data, simulated sensor noise could be added to improve realism, but it is essential that the inverse and forward mechanization algorithms themselves do not add any additional noise because of a lack of duality; they should be perfectly consistent with each other. This tutorial details the set of equations for inverse and forward mechanization. It also shows how to calculate velocity information from position information and how to estimate attitude information from velocities. As a demonstration of the accuracy of the equations, real-world pose and sensor data are used as inputs to the algorithms and the outputs are compared.