Underwater Vehicle Navigation Techniques Chris Barngrover CSE 237D
Most funding goes to UAVs followed by UGVs Lots of UUV applications (e.g. Moorea) GPS is easiest way to know location, but this fails underwater Need to use other techniques
Dead Reckoning Inertial Navigation System (INS) Doppler Velocity Log (DVL) Acoustic Techniques ◦ Long Baseline (LBL) ◦ Ultra-short Baseline (USBL) Geophysical (a priori maps) Computer Vision
Microstrain 3DM-GX1 INS SSI Technologies Pressure Sensor 2 Remote Ocean System CE-X-18 Underwater Cameras OpenCV Library
Convert pressure sensor data to depth Develop module that subscribes to INS, depth, and vision data Develop a Kalman filter to create position estimation Use vision techniques to rectify position estimation
Incorporated Planner Module Developed LPS Module Researched pressure to depth conversion Researched Kalman filter techniques
Depth Conversion Function Basic Kalman Filter ◦ Ground up development – Stalled ◦ OpenCV Libraray - Success
SSI Technologies Pressure Sensor Take depth measurements at DepthPSI (avg)PSI (mode)STDEV < <
Variables: Average Function: Mode Function: Amalgamation:
Created a kalman library ◦ init_kalman() ◦ close_kalman() ◦ kalman_update( time, status ) ◦ kalman_get_location( &loc ) Manages the CvKalman class from OpenCV
State Equation: : state vector : transition matrix - relates state vectors : control matrix – relates control to state : control vector : noise vector (k represents current time)
State Equation:
Measurement Equation: : measurement vector : relates state to measurement : state vector : noise vector (k represents current time)
Measurement Equation:
Continue Kalman Filter library ◦ Add control elements – ◦ Use angle and rotation angle to fix accelerations ◦ Add velocity sensor for better results ◦ Consider measured covariance matrices ◦ Use vision to rectify location ◦ Incorporate acoustic pinger triangulation Other related work ◦ Build standard course with dimensions ◦ Develop visual tool