Challenges for Autonomous Sailing Robots

Challenges for Autonomous Sailing Robots Ronny Eriksson, Åland University of Applied Sciences, Mariehamn/Finland, [email protected] Anna Friebe, Åland University of Applied Sciences, Mariehamn/Finland, [email protected] Abstract The purpose of this paper is to highlight challenges Åland Sailing Robots and other teams are likely to face during long distance sailings. Robotic sailing boats have to perform complex planning and manoeuvres to sail automatically and without human assistance. Sailing robots are affected by harsh environmental conditions. These challenges combined have, so far, made Atlantic Ocean passages impossible. Functional autonomous systems and management of the external environmental factors is a prerequisite for sailing robots in order to become widely used and commercially viable. 1. Sailing Robots - History and today's status 1.1. Rudder control Mechanical self-steering in the form of fixation of the rudder or tiller in a position has been used for a long time. The wind vane is a more advanced mechanical approach developed in the middle of the 20th century to keep the sailing boat at a constant course relative to the apparent wind. Electronic steering of the rudder began in the second half of the 19th century and became more common in the beginning of the 20th century. The first systems used a magnetic compass, and it was later replaced by the more reliable gyrocompass. Today’s systems often utilize GPS data and wind sensors in addition to a compass. Electronic rudder steering has often been based on control theory such as PIDregulators. Due to the dynamic environment of these systems, approaches such as Artificial Intelligence, Fuzzy Logic and Artificial Neural Networks have received considerable attention. 1.2. Sail control and design Automatic sail control, on the other hand, is much more recent. The first publications in this area are from around 1990, and there are still many remaining research questions in this area. Rigid wing sails have been compared to traditional fabric sails, and have several advantages. They can provide a better lift-to-drag ratio than fabric sails, are generally more reliable and avoid the problem of sail luffing or flapping. Drawbacks of rigid wing sails include difficulties to enable reliable reefing, and the pricing of lightweight, strong, rotating wing constructions. Rig designs have also been a focus of research, since a conventional sloop rig needs appreciable power to tighten the sails. A balanced rig design, Fig. 1, offers great potential in saving power. A common balanced rig consists of a mast with a mainsail and a jib. The mast passes through the boom, that extends forward of the mast. The sizes of the main and the jib are determined so that the combined centre of effort is just behind the mast. Most rigid wing sails are balanced. Sail control strategies in autonomous sailing are mainly focused on controlling the sail’s angle relative to the wind. Other components of sail control, such as mast raking, reefing, control of luffing and sail shape adjustment are for the most part areas of future improvement. Most sail control strategies applied for autonomous sailing robots are only based on the apparent wind and desired direction. Some systems include state machines for specific handling of manoeuvres such as tacking and jibing. Rigid wing sails are often self-trimming. In this case, a smaller tail wing is mounted just behind the main wing, and the tail enables control of the traction obtained from the wind. 67 Fig. 1: Example of a balanced rig, source: BalancedRig® https://www.balancedrigllc.com/ 1.3. Route control Route control is an area where strategies developed for other purposes may not be directly applicable for sailboats. The shortest route between two waypoints may not even be directly navigable depending on the wind direction. The optimal route depends on obstacles, the properties of the ship and weather conditions. Route control is divided into long term and short term routing. Long term routing can take into consideration weather predictions, sea charts and the boats properties. Short term routing is concerned with the navigation between given waypoints and for the most part deals with local sensor measurements only. Separation assurance and collision avoidance are other factors in route control. Methods can take into account weather and tides, AIS (Automatic Information System) and sensor data from for example radar, infrared sensor data and cameras. Considerable research challenges remain in this area and the strict energy requirements on long autonomous sailing journeys entails an additional difficulty. 2. Åland Sailing Robots – motivation and mission 2.1. Motivation Åland Islands have a long tradition of seafaring, and sailing is an important part of that heritage. Åland Sailing Robots is a possibility to combine the sailing tradition with modern invention and bring sailing to the future. At Åland University of Applied Sciences there are study programmes for Navigation, Marine Engineering, Electrical Engineering and Information Technology, among others. Åland Sailing Robots is an interdisciplinary project that ties the programmes and associated research together. In addition the project contributes to an extended international network. 2.2. Mission The goal of Åland Sailing Robots is to develop the first fully autonomous sailboat that successfully crosses the Atlantic Ocean. Through this project we will contribute to the research and knowledge on green technology and autonomous vessels. 68 3. The Microtransat Challenge – and outcome 3.1. Rules The Microtransat Challenge is a transatlantic race of fully autonomous sailing boats, no longer than 4 m (LWL). From 2015 there is also a class for boats where waypoints are updated remotely, where the system is not entirely autonomous. A class for motor boats or hybrid sail/ motor boats restricted to two meters is added in 2015. The 4 meter limit for sail boats will be decreased to 2.4 meters in 2017. Two routes are allowed, one West to East and one East to West. According to Microtransat’s interpretation of the International Rules for Prevention of Collisions at Sea, COLREGS, a vessel carries cargo or passengers. Therefore according to the Microtransat interpretation, the autonomous sailboats are not considered vessels, but buoys, and therefore they do not need to navigate according to COLREGS. 3.2. Attempts Since 2010 ten attempts have been made to cross the Atlantic, but none has been successful. In fact, all attempts have failed during the first ten days of the journey. There is no clear trend in the shortcomings, some boats have been lost at sea, others have ran aground, and there are examples that have been hit by another boat or caught in a fisherman’s net. Although this implies that autonomous sailing is an immature research area, there is an example, the Saildrone, that has sailed successfully in the Pacific Ocean from San Francisco to Hawaii. Also the most advanced, expensive and well-equipped boats may not be sent on a risky Atlantic journey, and that may affect the results. 4. Legal discussion and COLREGS 4.1. Criticism of the Microtransat interpretation The Microtransat interpretation that the sailboats are buoys has been criticized, since the sailboats could cause considerable damage to property and even harm humans in the case of a collision. In addition COLREGS state: “The word "vessel" includes every description of water craft, including nondisplacement craft, WIG craft and seaplanes, used or capable of being used as a means of transportation on water.” This means that a sailing robot is considered a vessel if it is capable of being used as a means of transportation, which is certainly true for the larger sailing robots. Legal expertise we have been in contact with assess that there is a reasonable chance that unmanned maritime systems will be considered vessels. Size, power, carriage of cargo or passengers, navigational capacity, ballast, lights, other equipment and registration are some of the factors that will be taken into account when determining whether the unmanned maritime system is a vessel. None of these are by themselves certain to predict the decision. Another thing to note is that these definitions may vary between different countries. 4.2. Need to recognize unmanned ships in regulations The existence of unmanned ships, particularly autonomous or semi-autonomous ships is not recognized in these rules. Adherence may not be strictly possible for unmanned and specifically autonomous ships until it is. Rule 5 on lookout, for example, states: “Every vessel shall at all times maintain a proper look-out by sight as well as by hearing as well as by all available means appropriate in the prevailing circumstances and conditions so as to make a full appraisal of the situation and of the risk of collision.” Since the rules specifically state human senses such as sight and hearing it is difficult to see that an unmanned system can oblige to this rule in a strict sense. One alternative would be to recognize autonomous ships in the regulations. Possibly the rules could allow them to navigate according to rule 19 for conduct of vessels in restricted visibility, at least under some circumstances. Specifically that would mean navigating according to rule 19 d) for avoiding other vessels detected by radar alone. An implementation of this rule would be an attempt to navigate according to COLREGS in a minimal but cautious manner. 69 Another issue is the fact that the Convention on Limitation of Liability for Maritime Claims (LLMC) may not apply to unmanned maritime systems. If unmanned and autonomous vessels are not recognized by law and they are operated at sea without a recognized code of practice, then it could be interpreted that a limit of liability might not apply. 5. Start up and funding of projects 5.1. Groups involved in robotic sailing Research on autonomous sailboats has been ongoing for about 25 years. Most projects are small, and based in the academia. The projects we are aware of are located in Europe and in North America. A community exists around the Microtransat Challenge, which has caused the spin-off World Robotic Sailing Championship. In the US, Sailbot arranges the International Robotic Sailing Regatta, which is a competition mainly for students. European groups working in robotic sailing include ENSTA Bretagne, France, Aberyswyth University, Wales, University of Porto, Portugal, University of Lübeck, Germany, Darmstadt University, Germany, NUI Galway, Ireland and Åland University of Applied Sciences, Finland. In USA work on sailing robots is performed at Cornell University, the US Naval Academy, Tufts University, Olin College among others. In Canada at least Memorial University and Queens University are working on autonomous sailboats. Saildrone Inc. is a business where this technology has been commercialized for marine research applications. 5.2. Funding A common feature of most of these groups is lack of substantial funding. Several of the projects are funded and developed by private individuals where the driving force is the person's enthusiasm. This implies that the development of sailing robots could go quicker than it has done until now. The main reason the development hasn’t gone faster is lack of commercial applications for sailing robots. Public research funding is therefore of utmost importance until commercial projects can fund research and development based on profit or expectation of profit. 5.3. Commercial use of sailing robots The demand for green technology such as wind powered boats may increase in the future. Also the decreased manning cost may enable new applications such as large-scale data collection. Autonomous vessels are suitable in hazardous environments. An autonomous sailboat with sufficient recharging capacities can be used in cost effective long-term missions. 5.3.1. Transportation It is likely that autonomous sailing robots can be used for transportation of goods in the future after some research and development. Such automatic transport system, without manning or fuel costs, could be commercially viable. However, the technology is not yet mature and juridical questions remain unresolved, hence it is impossible to start a commercial project at the moment. 5.3.2. Marine research Autonomous sailboats are suitable for automated data acquisition in the oceans. Sailboats can visit areas regularly and collect data of interest, such as salinity, chlorophyll, pH, dissolved oxygen, depth etc. Data can be transferred immediately. This can be of great help for oceanographic research. In marine mammal research, PAM (Passive acoustic monitoring) is often used to estimate the number of animals in an area. For certain animals, sailboats are a promising alternative. The first commercial application, the Saildrone, is in the area of marine research. 70 5.3.1. Surveillance Surveillance of borders or other areas of interest is also an application that is suitable for autonomous surface vehicles. Unmanned boats can also go into areas that may be dangerous for humans. 6. System components and challenges An autonomous sailboat integrates many different components. Development of an autonomous sailboat involves a number of competencies and there are several challenges that need to be addressed. It appears that almost every team lacks one or more of these competencies as per today. 6.1. Tracking of sailboats To be able to track the position of the autonomous sailboat is essential, since failure to do so means that the team will not be able to find and recover a non-functioning boat. A satellite based tracking system is most suitable for ocean passages. There are several commercial possibilities for satellite monitoring, such as Yellowbrick, Rock7, Spot, etc. These systems often use the low orbiting Iridium satellite system. Automatic Identification System (AIS) is a mandatory navigation safety communications system under the provisions of the Safety of Life at Sea (SOLAS) Conventions. AIS is an alternative that in addition contributes to making the autonomous sailboat more visible to other vessels. Satellite AIS tracking is a relatively new technology that has changed the landscape for monitoring the maritime domain. AIS signals can now be detected by a satellite in a low earth orbit and provide a global capability for monitoring all AIS-equipped vessels. Some projects have developed own radio communication system for short or medium range tracking. The team ENSTA Bretagne (and partly together with the Åland Sailing Robots team) inside the Microtransat community has developed a tracking system using 3G called SWARMON. It’s an embedded system to track marine robots in real time, mainly for testing and for presenting data from events like the World Robotic Sailing Championship. 6.2. Energy management For long term missions it is crucial not to use more energy than the system can generate. This means that the system will use energy efficient components and designs for electronics, motors and rudder control. It also means that the system may temporarily switch off parts of the system that are not necessary, and a need to generate sufficient energy through solar panels or other means. Other aspects of energy management are discussed in Section 1.2 on sail control and rig selection. 6.3. Robustness Ocean sailing is an extremely harsh environment that exposes sailboats to corrosion, water ingress, solar ultraviolet radiation and expansion / contraction due to temperature differences. At the same time, they are exposed to forces such as compressive and bending moments due to wind and waves. There are assorted issue to address: • • 71 Software Extensive testing of control software is necessary to ensure that it is working properly in all possible scenarios. It is also necessary to evaluate the memory management of the system and other aspects that may affect the stability in long term missions. Mechanics The mechanical components of the system need to be reliable in long-term use. They need to be able to withstand strain from water, salt, different temperatures and forces. Selection of suitable materials and robust designs are crucial. • Electronics Electronics components need to be safely protected from water. It needs to be ensured that they tolerate expected temperature intervals, and likely they need to be insulated to avoid extreme temperatures. It is an additional challenge to find electronic equipment that is reliable, and can withstand temperature changes while being energy efficient throughout the temperature interval. 6.4. Fouling An ocean crossing with a small sailing robot will take weeks or months as the sailing speed is low. During the sailing the hulls resistance penalties due to growth of slime, shell and weed. This phenomenon is well documented for larger and faster vessels but for small, slow-moving ships, this is an unexplored area. It’s, however, likely that small boats will be more affected and decrease the sailing speed relatively more. Selection of a suitable material for the hull is likely to have a large impact on the fouling. 6.5. Route planning Route planning includes manual planning before the journey, and manual or automatic adjustments of the route during the journey. Factors that need to be taken into account are sea-lane traffic, depth information, tides, weather and energy related factors. Route planning is complex with many variables that are difficult to estimate and predict. For example, some routes give more energy from solar panels, depending on hours of daylight and sunshine, solar angle and solar panel temperature. The shortest distance between two points is a great circle sailing and short distance gives less sailing time. Consequently, a deviation from the route, due to geography and weather factors to generate more energy may be counterproductive when it comes to distance and sailing time. Long term expeditions need to be planned for seasons when weather conditions are favourable. The north Atlantic is well known for strong winds during autumn and winter which could do a crossing impossible. On other hands, the trade wind at the equator is relatively favourable for an east-west crossing in autumn. Even when a route has been planned with normal seasonal conditions in mind, it is likely that harsh weather will jeopardize the journey. Long term routing strategies that take weather forecasts and information into account can reduce the likelihood of such an event. 6.6. Collision avoidance Depending on the size of the vessel and legal assessments, different strategies for interpretation of the vessels environment may be selected. As mentioned in Section 1.3 on Route Control, sensors may include AIS, radar, infrared and cameras. To consistently classify objects properly in the environment in different weather conditions is a very complex signal processing task. With increasing number of possible surrounding objects to take into account, the capacities of limited processing units may easily become insufficient. In addition it is not trivial to specify strategies for separation assurance and collision avoidance in environments with many surrounding objects even with correct object classification. This problem may be divided into several strategies. The first strategy may be separation assurance, to keep sufficient distance to other vessels, while navigating in accordance with COLREGS. For the case where this strategy fails, a strategy for collision avoidance should be considered. This strategy may not adhere to COLREGS. Minimizing the risk of collision is the most important objective in this case. In practice, a separation assurance strategy may not be very efficient, due to the attention unmanned vessels frequently attract from curious seafarers. 72 Other factors that decrease the risk of collision with other vessels are those affecting visibility. Lighting is one of these factors, colour selection of hull and sails also affect the visibility. Other factors include radar reflectors and the use of AIS as discussed in Section 6.1 on tracking. 7. Future Unmanned and autonomous vessels will be increasingly more common, as will unmanned and autonomous road and air vehicles. It is difficult to predict the speed of these changes, since they are dependent on many different factors. These include technical development in different areas related to energy, materials, control, signal processing, communications and more. Legal and regulatory factors are also important and affect the speed of this development substantially. Whether or not autonomous sailing vessels will be an important part of the future fleet is less clear. The influencing factors here regard sail control and design, and the efficiency of the wind propelled vessels compared to other means of propulsion. Probably hybrid solutions incorporating wind propulsion in combination with other solutions will be more successful. Fuel costs and the efficiency of alternative energy generation methods such as solar or wave energy in comparison with the sailboats will also affect the demand for these solutions. Factors such as ability to find resources for research and development are also of greatest importance. 8. Conclusions The development of autonomous sailboats requires competencies in many fields, and contains a number of research challenges. This is an inspiring and challenging area where finding the right resources and collaborations will be crucial for success. No matter if autonomous sailboats will be a common element of future seafaring, we are convinced that the work in this area will bring technology and knowledge forward. The efforts made during development of autonomous sailboats will contribute to increased interdisciplinary cooperation and greater knowledge on green technology and autonomous vessels. References CABRERA-GÁMEZ, J.; ISERN-GONZÁLEZ, J.; HERNÁNDEZ-SOSA, D.; DÓMINGUEZBRITO, A. C.; FERNÁNDEZ-PERDOMO, E.(2012), Optimization-based weather routing for sailboats, 5th Int. Robotic Sailing Conference, pp.23-33 DAHL, K.; BENGTSÉN, A.; WALLER, M.(2014), Power management strategies for an autonomous robotic sailboat, 7th Int. Robotic Sailing Conf., pp.47-55 STELZER, R. (2012), Autonomous Sailboat Navigation Novel Algorithms and Experimental Evaluation, PhD Thesis, De Montfort University, Leicester KLINCK, H.; STELZER, R.; JAFARMADAR, K.; MELLINGER, D.K. (2009), AAS Endurance: An autonomous acoustic sailboat for marine mammal research, Int. Robotic Sailing Conf. (IRSC), Matosinhos WARDEN, W.H. (1991), A control system model for autonomous sailboat, IEEE Proc. SOUTHEASTCON´91 Vol.2, pp.944-947 73