Asymmetric Dialogue

Marine robotics: developing autonomous marine systems

Dr Mae Seto, the Irving Shipbuilding Chair in Marine Engineering and Autonomous Systems, is working on intelligent autonomous marine systems, unmanned ships, shipboard intelligent launch and recovery systems, and marine robotics.

The Irving Shipbuilding Chair in Marine Engineering and Autonomous Systems will create a foundation for sustainable Canadian talent and leadership in marine engineering and provide leadership in Ocean Engineering at Dalhousie University. Irving Shipbuilding is contributing $500,000 (~€430,150) to establish and support the chair position, as part of their commitment to strengthen Canada’s marine industry under the National Shipbuilding Strategy.
Associate Professor Mae Seto, from Dalhousie’s Faculty of Engineering, has been appointed the Irving Shipbuilding Chair in Marine Engineering and Autonomous Systems. Work is already underway on many projects led by her through her Intelligent Systems Laboratory (ISL) in the Dalhousie University Ocean Technology Hub. Seto worked for 16 years as a defence scientist with Defence Research and Development Canada. Her experience positions her well to not only undertake this highly specialised research, but also support students in her lab to follow in her footsteps.

Interdisciplinary approach

Seto, and the team of students she supervises, work on intelligent autonomous systems, unmanned ships, shipboard intelligent launch and recovery systems, and marine robotics. Marine robots include unmanned underwater vehicles, unmanned surface vehicles and unmanned aerial vehicles (UAV). The research investigates fundamental development for scientific, military and commercial applications for robots.
The Intelligent Systems Laboratory is interdisciplinary and draws on the Faculties of Engineering (the Mechanical Engineering Department, and the Electrical and Computer Engineering Department) and Computer Science. The ISL specialises in areas pertinent to autonomous systems. Such systems can be stationary sensors (submerged arrays, intelligent nodes) or mobile robots. The ISL research programme highlighted here is in persistent maritime autonomy. Maritime here refers to underwater, on the water and well above the water.

The significance of the work

On-board autonomy is important and necessary to realise the value of operating robots in harsh and dynamic environments, like the oceans, where the robot is unable to communicate with its human collaborators very well (if at all). Autonomy facilitates the robot operating at more than arm’s length from the operator for extended periods and enables the robot to adapt to its own evolving state, the environment and its mission goals. Towards that, the ISL Persistent Maritime Autonomy Programme develops autonomous capabilities in the areas of:
  • Decision making;
  • Vision and learning applied to target recognition;
  • Mission planning;
  • Path-planning;
  • Robotic collaboration;
  • Tracking mobile targets;
  • Fault tolerance;
  • Underwater navigation (simultaneous localisation and mapping, terrain-based navigation);
  • Validation and verification
  • Underwater communications; and
  • Adaptive networking.
The Autonomy Programme is in an emerging area that is transformational in its impact on the complex roles that marine robots could perform. The architects and developers of the autonomy come from different fields.

Interdisciplinary students

It is necessary to draw and engage students from different disciplines to come up with creative autonomy solutions. The electrical and computer engineers provide expertise in signal processing, controls, communications and networking. The mechanical engineers have insight into the robot’s dynamics and kinematics, sensing, powering, propulsion, actuation and controls.
The advent of economical, small, low-powered, yet computationally powerful, central processing units (CPU) means the CPU could be embedded within a mobile autonomous robot and give the robot considerable on-board processing. Sensor measurements from marine robots could be processed, interpreted and exploited in situ towards better realising the mission goals.
An example of this is adaptive sampling to determine the source of sulphides to better localise a hydrothermal vent. In the past, scripted surveys would be performed in a suspected area with mixed results. The embedded processor with the added autonomy makes it possible to localise that vent, more directly and thus more quickly. Given powerful embedded processors are now integral components on-board advanced robots, the strengths of computer science students in the areas of intelligent control architectures, artificial intelligence, machine learning, vision and algorithm development are required.
Oceanography students (along with ISL external military and commercial partners) drive the requirements for developing autonomous ocean observation robots as end-users. They apply marine robots to make measurements that were not previously possible (e.g. intelligent sampling of nitrites at depth with unmanned underwater vehicles or the lower atmosphere for particulates with unmanned aerial vehicles) and are possible now due to the marine robots’ autonomy, scalability, and portability.
Given the complexity and range of skills required for creative autonomy solutions, ISL students from all these disciplines work together. This interdisciplinary approach has been crucial to the success of the lab and trains students that are better equipped to contribute as modern roboticists in industry.
One of the lab’s strengths is its industrial engagement which benefits all involved.

Industrial Engagement

Students within the ISL are encouraged to work closely with the lab’s industrial partners. This is developed and nurtured through on-going long-term relationships. Partners have inspired and/or procured funding for collaborative projects that introduce the partner to emerging technologies and tools that add to their portfolio and enhances their competitiveness. In the process, the ISL performs interesting academic and applied research that has an application and end-user. A collateral benefit for the lab is the opportunity for the graduate students to work with the partners through on-site internships. Such internships enrich the graduate student experience. It exposes students to real-world engineering at the research level and introduces the students to the partners for consideration as potential long-term employees. Through this lab-industrial partner relationship, the partner can also access services, capabilities and tools that are costly and of less value for the partner to maintain.
For example, a partner’s project requires access to advanced computational fluid dynamics (CFD) tools to validate the design for an underwater propeller duct. The partner does not require this frequently enough to maintain an employee in this highly specialised area or to pay for the high cost of the advanced CFD software licenses and computers. Industrial partners could potentially access these resources at the Lab through a services contract. Industrial partners could also propose undergraduate final year capstone projects to initially vet a concept with a small investment and very little risk to them. The ISL has hosted several such projects with good results. The ISL has several industrial partners that engage the ISL simultaneously at the undergraduate, graduate and services level. The results have been truly mutually beneficial for all involved.
The Irving Research Chair provides student stipends, support equipment and access to facilities for graduate research projects in marine engineering that have elements of autonomy. Two such projects are highlighted below.

Autonomous landing of quadrotors on ships in a sea state

The distance to the horizon varies with altitude and can be roughly approximated as: horizon (km)=3.57√altitude (m) . The distance to the horizon is nominally 5-6 km depending on the exact altitude above sea level on a ship. At a modest UAV altitude of 60m, the horizon has increased 3.5× relative to sea level.
On-board the UAV, the autonomous landing methods use acoustic and electro-optic sensors, and an inertial measurement unit with state estimation to enable UAV landing under poor visibility conditions (fog, mist) including night time. The UAV processor fuses this with the information transmitted to it from the ship’s motions through an inertial measurement unit that is under the UAV landing platform on the ship. Robotic collaboration, optimisation and learning algorithms and methods are applied to the sensor fusion to autonomously land the UAV. The objective is to land with no damage to the UAV or its sensor payload.
The work developed to date is undergoing testing and validation with a state-of-the-art motion capture system to provide ground truth UAV positioning and a Stewart-Gough platform which simulates the motions of the ship transom in a sea state. A variety of quadrotors, with embedded processors, are used in the testing.

Applications of autonomous UAV landings

The ability to autonomously land UAVs on the aft end of a ship that is surging, pitching, heaving, rolling, etc. enables it to work unattended for extended durations and facilitates new ways to use UAVs. The objective of such autonomous landings is that it is less invasive to ship operations, lowers risk to the ship and facilitates continuous UAV operation.
The applications discussed include:
  • Lower atmosphere sampling;
  • Identification of ship navigation hazards; and
  • Autonomous maritime search and rescue.
One application project is to sample the lower atmosphere along a specific track. For this case, the UAV payload sensor is an in-air sensor for monitoring lower atmosphere particulates. As a ship transits from one point to another, the UAV would take-off, become aloft, make measurements at altitude and then autonomously land back on the ship to download its data and re-charge its batteries for the next set of measurements. Several UAVs are used so no time is lost in this transition. When this system is installed on a ship that regularly transits back and forth between two ports (e.g. Halifax, Canada, and London, UK) then a unique set of scientific measurements could be made over an extended period at various times of the year and over several years.
Autonomous UAV landing on a ship in a sea state facilitates the use of UAVs for maritime searches. The autonomous landing methodologies are also applied for extended missions like constant situational awareness for the ship navigating through waters with navigation hazards like ice or surface clutter that are apparent from the air. It can also be used for surveys to track and map moving ice in open waters. Several projects looks at this using dynamic pose graph simultaneous localisation and mapping.
Autonomous UAV landings could also be applied to autonomous maritime search and rescue missions. An example of underway research and development follows.

Autonomous maritime search and rescue

The Chair is working with a local Nova Scotia company, Deep Vision Inc. (http://www.deepvision.ca/app-msat) and Hanseatic Aviation Solutions (https://www.hanseatic-avs.de/) in Bremen, Germany to develop intelligent capability on-board unmanned aerial vehicles for autonomous search and rescue applications in maritime environments. The increased horizon possible from an UAV at altitude increases the search rate by many times. The HAS S-180 and S-360 UAVs have proven capabilities in autonomous maritime search and rescue and are used as testbeds for ISL research.
Deep Vision Inc. develops advanced machine perception (software) tools that target maritime search and rescue which are integrated on-board UAVs. This trans-Atlantic Germany-Canada collaboration will create advanced autonomous UAVs for advanced civil, defence, security and search and rescue applications. This model of academic and industrial collaborations draws on the complementary research tools, people and capabilities at universities and the availability of platforms and professional engineers to perform the integration work. All parties participate in the testing and validation.
scitecheuropa.eu