Underwater Autonomous Vehicles Combine Robotics and Vision to Inspect Oil Pipelines
- Created: Sunday, 01 April 2012
The AURI incorporates many sensing techniques, such as ultrasound and Xray. The current prototype is equipped with visual inspection cameras. The four cameras use electrical signals controlled by a National Instruments Compact Vision System with synchronized LEDs. The lighting system is important in the underwater environment. The LEDs are on only during a brief period while synchronized with the camera valves, so they do not waste energy and can achieve an image with appropriate exposure.
As the vehicle may inadvertently rotate around the riser during the mission, a digital compass was added for acquiring exactly positioned images. The compass is connected to the vision system through an RS-232 connection.
The pressure transducer provides the computer with current depth information. The maximum depth is one parameter defined before starting the mission. Due to the uncertainty of the pressure sensor measurement (about 5 m), the depth value does not reference the images. For this function, two odometers were installed in two of the eight AURI wheels. Using two odometers ensures precise distance measurement even if one of the odometers slides.
The odometers have dual functions: to command superposition between shots, ensuring 100 percent riser coverage; and to provide closed-loop feedback to control tool speed. Using odometer feedback, the onboard computer adjusts propeller power to maintain a constant speed during the mission. The odometers feature digital serial peripheral interfaces (SPI) and connect to the vision system through digital port I/O.
Sensors installed in vessels with different pressure values monitor the temperature so the batteries can be recharged without opening the vessels, which could jeopardize impermeability.
The AURI is driven by two electric propellers, each with 1,350W, that together provide about 56 kg/f. The propeller must overcome the positive thrust of 10 kg/f and the forces of friction (dynamic, viscous, and cinematic) simulated with the data obtained in pool tests. The maximum submersion speed is scheduled at around 0.5 m/s and the immersion around1.2 m/s.
The whole system is powered by two sets of batteries. One is used for all control systems, computers, and sensors, and the second powers the electric thrusters. Once the vehicle is placed in the water, the only human interface with the system is a magnetic switch located inside the pressure vessel with the onboard computer. A small Hall sensor positioned in a specific place was used. A diver must deploy a magnet in this place for three seconds to start the mission. This boot mechanism does not have movable parts, does not require opening any pressure vessel, and is strong enough to prevent undue drives.
The vision system is the center of this intelligent tool, managing all tasks related to the camera, acquisition sensor, thruster control, compacting, and IEEE 1394 80-GB hard-disk data storage inspection. It also provides an Ethernet connection for software updates, on-the-fly sensor and actuator test capacity, mission parameter definition, and data download. A small Wi-Fi antenna helps connect the vision system so that when the AURI is on the surface, the user can control it without cables.
The control software, developed in LabVIEW, is responsible for various tasks, including sensor acquisition, camera command, data storage and occurrence record inspection, IEEE 1394 barring management, vehicle return command, closed-loop speed control implementation, and interface coordination through a TCP/IP terminal in the vision system real-time environment.
Control algorithm development and AURI post-processing were advanced even before the electromechanical project was finalized by implementing simulators using the available electronic hardware, including the vision system, cameras, lighting, storage units, and propellants, that were purchased at the beginning of the project. Thus, a simulation and test system was assembled using hardware in the loop and simulated a complete riser inspection mission.
Analytically, the hydrostatic and hydrodynamic model was based on the project rotary that was developed. This way, the whole system can be tested for performance, IEEE 1394 barring configurations, camera and lighting shot synchronization, thruster and acquisition command, and sensor data storage. The controller can be tuned in speed closed-loop in advance with the simulation data. When the field tests started, only a few adjustments had to be made. The programs for acquired data post-processing and result presentation were developed using LabVIEW system design software.
Important measurements were taken during pool tests to validate the AURI computational hydrostatic and hydrodynamic models. By configuring the AURI program to acquire the sensor signals at higher rates than necessary for normal inspection, the system friction coefficients can be raised, including the viscous friction of water passage through the pipeline and the structure, and the dynamic roller suspension friction. The measurements confirmed that the expected buoyancy would be 10 kg and the hydrostatic return emergency (without the thrusters) would be 0.6 m/s. The mission to inspect 1,000 m of the riser took 44 minutes, with a significant safety margin on the available energy.
In the field, the initialization sequence performs all peripheral and sensor tests in approximately one minute. Then a light flashes to inform the operator that the AURI is ready to start the mission. At this point, the diver moves away from the vehicle because the thrusters soon activate. When control software performs the mission, the AURI automatically connects to return. Although the thrust is positive, the AURI may revert the propellants to overcome any obstacle.
With LabVIEW, the AURI was programmed for each mission’s special features, and the same programming language was used for post-processing to transform the acquired data into useful information, showing mission results and hardwarein- the-loop simulation platforms. The vision system FTP and TCP capacities provided a simple way to transfer all acquired data and customize mission parameters through any computer or PDA. AURI remote software virtual instrument access provided a way to diagnose prototype failures.
This article was written by Ricardo Artigas Langer of EngeMOVI, Brazil. For more information on the National Instruments hardware and software used, visit http://info.hotims.com/40432-320.