Underwater Vehicles Technology Advances

[link—standalone]

An Introduction to AUVs technologies

Introduction

Autonomous underwater robots have become an integral part of both marine science and ocean engineering, and with the evolution of these sectors, the development of AUVs is heading towards full autonomy. A complete autonomous underwater vehicle will operate independently and will be able to act logically and methodically to accomplish the required task. Over the past years, many attempts have been made to apply autonomy on AUVs in the form of surveillance and data collection, yet a fully autonomous vehicle that will be able to act intelligently within the operating environment has not been achieved (Ridao et al., 2015; Sahoo et al., 2019).

From the beginning of the 90s, various projects focused on different aspects of underwater autonomy. One of the first attempts was made with the autonomous vehicle ODIN, which was equipped with a one degree of freedom robotic arm and was capable of operating either autonomously through an onboard computer or via a tethered cable (Choi et al., 1994). Another attempt towards autonomous vehicles was the OTTER project conducted by MBARI’s research programme in 1995. The vehicle was equipped with a robotic arm, cameras for object detection and a variety of sensors including depth sensors, and compass. OTTER’s mission was to retrieve an object from the seabed and it was one of the first vehicles with computer vision, and interestingly for the first time, the idea of “intelligent control” for an underwater robot and the design of a complete autonomous vehicle was presented (Wang et al., 1995). Another significant project which developed in that period was the AMADEUS subsea robot. The robot was designed to handle and grasp underwater objects and was equipped with a pair of three-fingered hydraulic robotic arms. The double robotic arm was innovative for the time as there was cooperation between the two arms for object manipulation and with the use of the three-finger gripper, it would be possible to control the pressure when grasping an object, which is significant for precise gripping (Lane et al., 1997). Although different projects have been developed over the past years with the focus on creating a fully autonomous underwater vehicle, there has not yet been a vehicle which will intelligently interact within its operating environment. Therefore, the current project will develop a system capable of critically acting for underwater intervention and fill the gap of previous research.

AUV State-of-the-art

During the last two decades, different projects started to develop, and the focus was to create an AUV capable of intervention. These projects include the EU-Thermie funded SWIMMER, the ALIVE project, the SAUVIM project, the TRIDENT I-AUV project, the GIRONA 500 and finally the MARIS project. The SWIMMER vehicle was an AUV which was capable of transporting a work- class ROV to a dock station at the seabed. Once the AUV autonomously docked to the underwater platform, the ROV was deployed to execute the operation. The ROV was connected with a fibre optic cable to the SWIMMER AUV, which was deployed to perform its mission, to dock in a predefined underwater platform. With this configuration, there is no need for a support vessel with the necessary equipment to deploy and operate the ROV (Evans et al., 2001). Even though the earlier projects, SWIMMER, OTTER and AMADEUS, were pioneering for their era, they were still dependent on human supervision and guidance to perform the given task; this was revolutionised with the ALIVE project.

The ALIVE project was a breakthrough for underwater intervention missions due to its abilities in conducting physical manipulations while hovering to keep its position. The AUV was fitted with a seven degrees of freedom robotic arm and was capable of performing autonomous docking. The docking prcedure was performed based on the data received from sonars and cameras, which allowed the robot to estimate its position at any time. Once the docking had been achieved, the robotic arm was deployed to perform the preprogrammed task, which was the opening and closing of a valve in an underwater panel (Evans et al., 2003).

The advances in underwater robotics were continued with the SAUVIM project which was funded by the University of Hawaii. The Hawaiian project was conducted over the same period as the ALIVE project but followed a different approach for underwater intervention. The vehicle was designed to perform the manipulation task while hovering at the target area and to achieve the task, three different localisation approaches were used. Firstly, a sonar was used to perform long-distance searching to find the location of the desired target. Secondly, a different sonar was used for object identification to set the underwater robot to a specific position for manipulation. Finally, when the robot was at the manipulation position, an ultrasonic camera was used to perform the actual intervention. The robot was fitted with a seven degrees of freedom arm with a camera on its palm. During the tests, the robot performance was deemed satisfactory since it was able to identify and recover the target object (Marani et al., 2009).

In 2010 a new approach to underwater intervention was proposed with TRIDENT, an EU funded project. This project proposed flexible underwater operations using two different vehicles, one a surface vehicle and the second as the Intervention AUV (I- AUV) which initially performs mapping of the area of interest then returns to transmit the data to the operator. Then, the operator decides on the desired task and the interventions, and the vehicle descends again to perform the instructed mission. The TRIDENT project was tested in an underwater environment and assessed for its cooperation capability between the vehicles and for intervention skills of the I-AUV to execute the recovery action (Ridao et al., 2015; Sanz et al., 2012).

The TRIDENT project was used to host another newly developed AUV capable of intervention, the GIRONA 500, which was developed by the Spanish University of Girona in 2012. The underwater vehicle was designed to operate in depths of up to 500m and to be as reconfigurable as possible. The vehicle’s capacity to carry a variety of equipment and instruments depending on its mission was a distinct characteristic of the particular project. Also, the AUV was fitted with a range of thrusters where, in its most simple setup, it used three of them to control the vertical and the horizontal motion of the vehicle. In contrast, it could be fitted with eight thrusters which gave the vehicle the capability to operate in all six degrees of freedom for intervention tasks (Ribas et al., 2012).

GIRONA 500 is controlled using two computers, one is a low power computer which is used for the main tasks of the robot, and the second computer is more powerful and is dedicated for computational demand processes such as mapping and image processing. The second computer is operated only when it is needed; thus, power saving can be achieved. The software architecture implemented on the robot is based on three different operational layers which may exist independently. One layer is used for mission planning which can be done autonomously or by being given instructions through a language named Mission Controlled Language (MCL) and is executed under the vehicles software architecture, named Component Oriented Layer-Based Architecture for Autonomy (COLA2). The next layer is used to control the robotic arm, the execution layer, through cameras, sensors and actuators. The precise position of the arm can be set according to the target’s position using Petri Nets to perform the manipulation task. A full description of Petri Nets is given by (Palomeras et al., 2008). Lastly, the final layer is used to coordinate the mission and the execution layers. The coordination layer is referred to as the reactive layer and can execute a variety of tasks from the most simple such as starting a camera and different sensors, to the more advanced and complex such as 3-D navigation and object manipulation (Ribas et al., 2011).

The capabilities of GIRONA 500 were tested in an underwater environment where different tasks were given to the vehicle such as autonomous docking, valve manipulation and object recovery. Notably, the task of turning a valve was a milestone for underwater intervention since the robot had to perform the turning operation based on the Learning by Demonstration technique (LbD), also known as Programming by Demonstration (PbD) (Carrera et al., 2014). The LbD technique is based on the idea that the learning process can be achieved in an automatic programming method where the user presents or demonstrates, the new action or behaviour that the robot should learn rather than programming it through coding (Mohammad & Nishida, 2015).

Finally, the Italian project MARIS developed with a focus on manipulation mainly for the offshore industry as well as for a variety of scientific tasks such as ocean surveying. The underwater robot uses a combination of sensors for navigation such as Doppler Velocity Loggers (DVL) and acoustic methods to be able to orient itself with reference to the seabed. The robot can detect objects using stereovision technology to generate a more accurate 3-D interpretation of the object. Moreover, it uses a sophisticated robotic arm which has been designed to intelligently perform grasping using an integrated system of force- torque to “sense” the object. Another feature of the MARIS project is that the operation can be performed using one or more vehicles, where they can coordinate to execute the grasping of the object and then to transport it with safety for the object and the vehicle (Casalino et al., 2016).

Summary

A variety of projects have been developed with the focus of underwater intervention. All projects have developed AUVs that are capable of autonomous data collection using a variety of onboard sensors, and also have the ability of physical intervention by performing pre-programmed tasks.

References