Development of a real-time aided inertia navigation system for an underwater robotic vehicle [thesis]

Dzulkifli Mohyi Hapipi Mohammad
2005 7*0 Q£ ATTENTION: The Singapore ABSTRACT Underwater robotic vehicles (URV) have a practical significance in the offshore engineering field, especially in underwater environment surveying, man-made structure inspection, and pipe and cable laying. Most URVs are designed to be autonomous as in autonomous underwater vehicle (AUV) or remotely operated vehicles (ROV). Modern URVs uses onboard navigation system to provide the necessary information about its environment. It is one of the primary
more » ... ne of the primary components in all types of URV and it is used to define the states of the vehicle such as acceleration, velocity, position and angular rate and position. It consists of the physical hardware such as sensors and the computer system and the software layer such as the data acquisition, data management and the navigational algorithms. This thesis explores the development on an Aided Inertial Navigation System (AINS) for a compact, low cost twin-barrel URV developed by the Mechanical and Production Engineering (MPE) department of Nanyang Technological University (NTU), Singapore. This system uses low cost sensors, which produce large drifts and bias errors. Low cost sensors have errors up to 1000 times larger than high cost precision sensors. The URV utilises the IMU600CA-200 inertial sensor from Crossbow and is aided by non-inertial sensors such as Argonaut MD Velocity Doppler from Sontek, KVH compass, Tritech Altimeter, Crossbow Tilt sensor, Sonavision scanning sonar and Falmouth pressure sensor. Relying of the fast update rate of the IMU, the additional sensors aids the IMU by correct the vehicle states at periodic interval. All the data processing had to be done by an onboard Octagon microcomputer running at 133 MHz. For optimal performance, QNX 4.25 operating system was used to enable real time processing for the sensors. For the navigation system to function as required, the sensors, computer hardware and operating system need to be optimally configured. Data acquisition and data management architecture were designed for the system. One of the primary factors that affect the system performance is in obtaining data from the sensors. A polling only configuration requires the process to wait for the data after the poll thus results in time wastage. On the other hand, interrupt only configuration requires the computer to service the interrupt, thus affects the running process. It was observed that using the interrupt/polling combination provides the best performance to the system since the process releases the CPU for other processes under idle states. A shared memory objects was used to manage the data, as it requires minimum synchronisation and allow other processes to extract it when required. It also complements the network data transfer via both message passing and message queue via Ethernet link between the URV control and navigation pod. Navigation algorithms such as frame transformations, kalman and low pass filter, software integration and gravity and steady state compensation were designed and tested for the system. These algorithm, affects the performance of the designed architecture which is fundamental in maintaining a real time acquisition and processing. Therefore, the algorithms were tested based on the navigational accuracy and processing speed. Lab and navigation experiments were carried out to test the robustness of the algorithm which is represented by the accuracy of the navigational performance. A stable, robust real-time architecture is fundamental in the development of the navigation system. ii ATTENTION: The Singapore Copyright Act applies to the use of this document. Nanyang Technological University Library ACKNOWLEDGEMENTS
doi:10.32657/10356/6166 fatcat:4mizimbptfhpdh35r2uwt2kdfi