Module imu mod IOISA_v 0.1

Overview

This IMU related module is designed to perform IMU initial earth self-alignment for camera attached frame. This self-alignment is then typically considered as initial conditions for camera attached frame integration over time. The alignment is performed considering the longest detected inertial still range in IMU sensor measures.

Module nomenclature

This stream module takes three input streams : two input physical streams that stores IMU angular velocities and translational accelerations and an unitary stream that contains detected inertial still ranges boundaries. The following image summarize the module nomenclature :

The three output physical streams store each the three components of one unit vector that defines earth aligned frame initial conditions. The unit vectors components are stored twice to gives low-timestamp and high-timestamp on which initial conditions apply.

Module interface

This section gives a summary of the module interface in terms of streams and devices. The order of devices and streams is given as waited by the module. The devices interface is given by :

device( Physical ) IMU sensor

The stream input interface is :

stream( Physical        | IMU ) Gyroscopic velocities
stream( Physical        | IMU ) Translational accelerations
stream( Integer unitary | IMU ) Inertial still ranges definitions

and the stream output interface is :

stream( Physical | IMU ) Frame initial x-vector
stream( Physical | IMU ) Frame initial y-vector
stream( Physical | IMU ) Frame initial z-vector

The two first input streams are attached as well as the three output streams.

Module operation

The earth self-alignment procedure is based on detected inertial still ranges. It typically considers the longest detected range and compute the mean of the accelerations and angular velocities provided by the sensors, assuming only gravity and earth rate contribute to sensor measures.

The alignment process starts by computing the rotation matrix that brings the z = (0,0,1) vector on measured anit-gravity vector in order to have z aligned against gravity. The rotation matrix is then applied on the frame unit vectors.

The mean angular velocity vector is them converted in the rotate frame using the previous rotation matrix. The second part of the alignment consists to determine the rotation around the z axis of the anti-gravity-aligned frame needed to bring the (x, y) projection of the mean angular velocity in the (y,z) plane of the previous frame