Andrew Dykes reports on the winners of Total’s ARGOS Challenge, which sought to find and develop the first surface-based autonomous inspection robot for oil and gas sites
Subsea work is ruled by fleets of ROVs and AUVs. Advances in robotics and communications have propelled these machines to their current status as the indispensable workhorses of the offshore and marine energy industry. In doing so, operations which might have previously taken weeks of dangerous work by diving teams can be accomplished in hours by a few well-tooled robots. Why, then, had we not done the same for surface work? Even as the industry moves towards the long promised digital oilfield, installations tend to be maintained and inspected by personnel, even in cases of extreme isolation or environmental conditions. But with the right tools, remote and autonomous robots could monitor and maintain these installations, lowering the amount of human intervention required, improving safety and saving considerable operating costs.
The initial problem, in the eyes of supermajor Total, was that this technology did not yet exist. Speaking by phone, Total E&P prospective and R&D strategy senior manager Sylvie Duflot told InnovOil: “At the time there were no other autonomous robots for oil and gas, so it wasn’t possible to [invest in one directly]. It was our intention to really be the ones to create the first autonomous surface robot for oil and gas sites.”
It was with this goal in mind that Total launched the ARGOS (Autonomous Robot for Gas and Oil Sites) Challenge in December 2013, in partnership with the French National Research Agency (ANR). Putting out an open call to international universities, OEMs and labs, ARGOS sought designs for the first autonomous surface robots adapted specifically for oil and gas sites, and compliant with ATEX/IECEx (explosive atmosphere) standards. Teams of engineers would then pit their robots against each other in series of scenarios and tests in pursuit of a 500,000 euro (US$560,000) prize.
This would allow Total and its judges, assembled with the help of ANR, to survey the state-of-the-art equipment in terms of robotic capabilities. Time was also of the essence if the company was to be serious about deploying these vehicles to real assets, Duflot said. “We thought an open innovation challenge was the best way to get the best team and to speed up the R&D process.” A total of 31 consortia submitted applications, with five teams selected to compete. Each was given up to 600,000 euros (US$670,000) in funding from Total, and following a launch meeting in September 2014, each group then prepared for the first round of trials the following summer. After a long, trying road and three difficult rounds of tests, the winners were finally announced in May 2017.
Intelligent design For Total and the ARGOS judges, it was important that these tests were not easy. If robots are to be deployed in these environments, they must be capable of dealing with a wide variety of conditions and tasks without physical intervention. In addition, unlike subsea ROVs these robots do not have the benefit of a tether system for communications and power.
In its challenge brief, Total noted the problems of avoiding obstacles; climbing stairs; taking equipment and environmental readings; working in poor weather conditions (e.g. wet or slippery floors) and detecting anomalies such as fluid or gas leaks. Moreover, each trial presented the teams with increasing levels of difficulty. Many of these might be mitigated by good design and skilled operation, but the robots also had to be capable of operating autonomously in a number of tough situations. ARGOS required them to perform some inspection unassisted, including carrying out “spot” reporting and inspecting hard-to-access or isolated areas.
In addition to autonomy, the robots had to be intelligent. ARGOS Challenge project manager in R&D Kris Kydd noted: “The robot must be able not only to read and record values on instruments, but also to autonomously determine whether or not the values it reads are within a normal operating range.” The teams were then tested on how their innovations handled emergency scenarios – for example, being able to send an emergency signal to operators when an oil or gas leak occurred or when those readings were abnormal, and to transmit data and images of the situation in real-time.
The site of the trials was a decommissioned gas installation in Lacq, southwest France. The UMAD (Availability Unit) was a former gas dehydration unit managed by SOBEGI and is now used for training human operators and emergency response teams. What better place to put the teams to the test?
The final five The final teams selected to compete included Japan’s AIR-K, Spain and Portugal’s FOXIRIS, France’s VIKINGS, LIO from Switzerland and the Austro-German ARGONAUTS.
While the competing designs had some common elements – four teams opted for tracked crawlers, and most for a pivoting sensor on top for imaging and detection – the robots were all fairly distinctive, and a far cry from the box-like homogeneity of subsea ROVs. Designing far outside the conventional box, LIO even opted for a legged quadruped with sensors it in its feet, truly a step away the crawlers that have dominated the robotics industry for decades. For the remaining four, stairs were to be navigated by movable flippers on which the tracks were mounted.
All adopted various levels of environmental proofing, including for dust and other debris, as well as waterproofing, enabling them to operate in the remote and extreme environments expected of them. Sensors also had to be robust – the housing used in the ARGONAUTS equipment, for example, was fully compliant with IP67 regulations. Some also opted for greater freedom of movement in their inspections. By the end of the competition, FOXIRIS had upgraded its design to include a full robotic arm with cameras and detection equipment.
Battle commences The first competition, held in June 2015, focused on navigation and inspection on the ground floor of the Lacq UMAD installation. However, even setting up this round this proved challenging, DuFlot said. “We needed a very robust communication system, and what we got in the first round was five different wireless systems for controlling robots. It was a difficult moment, I would say, but luckily we have some top telecommunications engineers in Total and we were able to hold the competition.”
Was Total aware at the difficulty of the challenge it had set the teams? “To be honest, no!” Duflot said. “I believe it was the same for the teams. It was a really difficult challenge because we really wanted the robot to have these requirements – autonomy, ATEX certification, for it to be reliable – so we had big expectations and the bar was quite high.” Less than a year later, in April 2016, the second round of challenges upped the difficulty. Duflot was present during the event and “was quite amazed by the improvement of the robots from the first stage in the competition. They were working a lot to improve their robots and all made incredible progress.”
The first day tested a robot’s ability to navigate a staircase autonomously, taking itself to the first floor of the structure and back unassisted. It also asked the robots to navigate an obstacle, assess its size and dimensions and then decide on whether that obstacle was passable, or alter its route. Most proved to be more than capable of independently tackling stairs – perhaps the mortal enemy of robots past – but over-sensitive instruments let down teams like LIO, whose robot detected obstacles correctly, but produced too many outliers.
The second day of testing forced the vehicles to search the site for a heat source. Here ARGONAUTS failed in its first run around the UMAD structure, but the robot moved fast enough to allow them another sweep, which was successful. A subsequent test saw the judges move gauges and instruments around the site. Robots then had to search for each and report back any inconsistencies between the installation and the 3-D model the robot used for reference.
The final day of tests threw everything at the teams, including water leaks, a loss of Wi-Fi communications, a simulated gas leak which they had to detect, and tested their ability to climb and descend stairs both autonomously and with manual control.
At the end of the three days, VIKINGS was in the lead.
By the third round in March 2017, some of the robots were unrecognisable from their original forms. Refinements, additions and work on ATEX certification produced vehicles which already looked field-capable. The latter requirement formed one of the first assessments, with ATEX experts brought in to confirm the compliance for the vehicles to work in a potentially explosive atmosphere.
Over the course of a week, robots were given new tasks, including navigating gravel, to detect mobile, traverse obstacles (barrels, flexible hoses and ditches) and to conduct acoustic monitoring autonomously using pump signature analysis. “During the last competition a member of the jury even stood in front of the robot without prior warning, testing its ability to stop and identify the obstacle before changing its route and mission plan,” Duflot added.
Building on the tests in the second round, the robots also carried out inspections and spot tasks in autonomous mode, and were assessed on how easy an operator was able to intervene and take control. One mission also included an emergency alarm to assess the robots’ response. The practicalities of operating these vehicles were assessed too. One additional requirement for the challenge was that the robot had to return to its docking station autonomously in order to recharge its batteries when it detected they were running low – perhaps one of the more simple tasks but one which would be vital to real-world operations. After a punishing week, the teams returned home and the judges began to deliberate their final verdict.
The winner is… On May 11, Total announced that the winning team was the Austro-German ARGONAUTS. Aside from their excellent choice of name, Duflot said that their robot had proved to be the most capable of handling everything the competition threw at it. “The competition had very clear rules and at the end of the day when we compared the position of all the robots, ARGONAUTS was really the winner. The robot exceeded the requirements of the Challenge rules by obtaining the ATEX certification prior to the third competition. They had the most advanced level of technology maturity. It was very robust and they had a well-engineered system.” For the team, there is no doubt that the competition presented a formidable challenge – speaking by email, ARGONAUTS told InnovOil that: “The mere complexity and difficult requirements made it most demanding project we have encountered in our professional robotic life.” ARGONAUTS had designed their system on a modular concept too, allowing for adaptations when confronted by future missions. Elements underneath the chassis also contributed to their victory – Duflot mentioned a very accurate positioning algorithm which allowed the robot a high degree of accuracy (to the centimetre) in navigation and performing readings. “The human-machine interface was really developed and easy to handle, and this is very important. It could also switch from one mode to another, from autonomous to remote, very efficiently,” she said. “All these points count toward a very reliable and robust robot. We need a good base to improve it and to move towards the robot of the future.”
What now? With a winning partner chosen, Total is now planning the next phase of development. Duflot said that the group would begin an industrial pilot at one of it sites by late 2017 or early 2018. “This will be a way to check how the robots work in the human environment. Our fields are developed for human [access], so it’s important to place the robot in a real site to be confident in its ability to work there. This is the next step for late this year or early next year.” ARGONAUTS added that externally, its design for the industrial pilot would not differ much from the challenge’s final version. “However, robustness, endurance and overall usability of the system will have to be greatly improved,” they said.
Duflot declined to say whether any particular installations were in the running, but outlined that a prerequisite would be an asset with a good wireless system and 4G connection. “Also, somewhere we can study the behaviour of the robots in the first year. From there it will be case by case,” she added. The results of this pilot will determine the group’s longer-term strategy with autonomous robots, but thanks to the progress made over the past three years of the competition, much of the basic capability testing has been completed. “Our intention now is to move fast and to be able to use these types of robot as fast as possible,” she affirmed.
As well as challenging them, the team has also developed a solid relationship with the supermajor. “Working with Total has been a great and enriching experience - after all it is a huge multinational oil and gas company and we are a consortium consisting of a small robotic company and an university institute,” they told InnovOil. “Still, we were always treated as equal partner in a successful project that will create mutual benefit for everybody involved.” For those eager to learn more, Total E&P and ARGONAUTS will be presenting findings from the experience at a number of conferences this year, and Duflot is confident that the industry will be interested in what they have learned. “Digitalisation is moving very fast and we have to think how to incorporate it and robotics at our sites,” she added. “We have some work to do, but clearly ARGOS is a very important first step, because the challenge provides us with very good information about what we can do with these autonomous robots.”