Elsevier

Ocean Engineering

Volume 36, Issue 1, January 2009, Pages 24-38
Ocean Engineering

Fòlaga: A low-cost autonomous underwater vehicle combining glider and AUV capabilities

https://doi.org/10.1016/j.oceaneng.2008.08.014Get rights and content

Abstract

The paper describes the current developments of a class of low-cost, light-weight autonomous underwater vehicles for coastal oceanographic applications; the vehicle class is named Fòlaga, the Italian name of an aquatic bird that swims on the water surface and dives to catch fish. The main design characteristics of the most recent vehicle of the class, the Fòlaga III, are reviewed. Navigation and control system design are discussed, with particular attention to the diving phase, which is accomplished as in oceanographic gliders by varying the vehicle buoyancy and attitude. Experimental results show that the PID robust controllers implemented are effective in the diving control phase. Finally, a distributed cooperation algorithm to be applied by a team of Fòlaga-like vehicles in adaptive oceanographic sampling applications is described. The algorithm optimizes area coverage while taking into account the accuracy in the reconstruction of the oceanographic field and inter-vehicle communication through a range constraint. The resulting dynamic programming algorithm can be implemented in a distributed fashion among the team components.

Introduction

Autonomous underwater vehicles (AUVs) are on their way to reach the technological maturity and the widespread diffusion enjoyed by their robotic predecessors, i.e. remotely operated vehicles (ROVs). However, the gap in number of applications, costs and market availability between the two classes of unmanned underwater vehicles has not been filled yet. Few years ago Alvarez et al. (2004) observed that the main limitations to the diffusion of AUVs in operational scenarios are those of vehicle cost and user-friendliness. The great majority of existing AUVs have been originated directly from research prototypes: their development was led by the general research interest of increasing the vehicle autonomy and not by specific mission needs. The absence of a well-defined mission goal at the design stage has favoured prototypes able to account for several potential missions, eventually leading to conservative design choices, as for navigational capabilities, depth ranges, allowable payloads, energy consumption and safety systems. This has in turn led to expensive final products and to operative mission procedures requiring the presence of a team of trained engineers, knowledgeable with the system design, in stark contrast with ROV operations.

In an attempt to reverse the trend, the authors have proposed in a series of papers (Alvarez et al., 2004, Alvarez et al., 2005; Bozzo et al., 2005) the design and subsequent realization and field test of an autonomous vehicle to be employed as a sensor platform for investigation of ocean mesoscale dynamics in shallow coastal waters. With design focus on the mission, a very simple and yet effective prototype has been realized, with great reduction in costs. The resulting vehicle has an actuation system combining those of autonomous gliders with those of propulsion-driven AUVs. In particular, the vehicle can carry a CTD package as payload; it navigates on the sea surface when in transit from one measuring station to another, and it submerges vertically when on station to perform the measurement. When on the surface, the vehicle has continuous GPS contact and land-station contact through a mobile phone link or a dedicated radio link. The land-station link allows for on-line modification of the mission requirements and for almost real-time data transmission. Due to this peculiar behaviour (navigate on the surface and dive when needed by the mission), and for its range of action limited to shallow water areas, the vehicle was named “Fòlaga”, the Italian name of the coot aquatic bird. Since there have been several evolution of the original design, we now indicate with the “Fòlaga” name the generic class of vehicles which respond to the above-specified behaviour.

Since Alvarez et al. (2004), several other research groups have proposed development of small-size, inexpensive AUVs with design focused on specific goals. Without the ambition to be exhaustive, the following developments need to be mentioned: Dunabin et al. (2005) have proposed a hybrid vehicle, mixing ROV and AUV capabilities, for coral reef exploration; Desa et al. (2007) describe an autonomous surface vehicle designed for measurement of water surface chlorophyll distribution, to provide calibration data to satellite remote sensing systems—the same group is also actively developing a small coastal water AUV Maurya et al. (2006) and Wood et al. (2007) have presented an autonomous surface vehicle, with autonomous mooring capabilities, for coastal meteorological and oceanographic data collection; the motion capabilities of the “Seabird” vehicle (Araki and Ishi, 2007) are similar to those of the Folaga, although diving is obtained through the combined action of orientable propellers and hydrodynamic surfaces, while the Folaga dives through internal change of buoyancy and attitude; however, the Seabird is intended for camera images collection in littoral waters and it is designed for a maximum diving depth of 10 m.

It is remarkable that all these developments, as well as the Fòlaga's, share the design methodology of starting from the mission objectives and then proceeding by carefully selecting a set of technical and methodological solutions specifically tailored to the mission needs. The result, in all cases, is indeed that of low-cost, light operation, adequate performance. It is also remarkable that the vehicles are very different from one another (in shape, navigation methodologies, instrumentation, energy sources, etc.), confirming what has already been witnessed in the ROV field evolution: mission-oriented design contributes to the diversification of the available AUV fleet, to efficiency with respect to the mission and to overall reduction in costs.

One additional possibility that immediately opens up as low-cost vehicles become available is that of using a team of vehicles to fulfil the mission needs. In this respect, it has to be mentioned the pioneering work of Wick and Stilwell (2001), whose low-cost, miniature AUV design was directed toward the research need of experimenting in the field of multi-vehicle cooperation and coordination strategies (Stilwell et al., 2004). Multi-vehicle or multi-robot cooperation/coordination, not necessarily referred to underwater applications, is presently a very active field of research, both from a methodological and an application-oriented point of view. However, the underwater environment poses severe limitations to the transpositions of methods originally developed for terrestrial or aerial robotics, due to the peculiarities of navigation and communication constraints. Recent examples and developments in underwater vehicle coordination can be found in Leonard et al. (2007) and references therein, in the context of oceanographic sampling; other relevant discussions can be found in Ghabcheloo et al. (2006) or Porfiri et al. (2007) and references therein. In this context, most of the underwater research, to the authors knowledge, has been directed toward coordination and formation control problems. As an exception, Caiti et al. (2007) focused on cooperation of AUVs, and in particular in distributed on-line planning of sampling stations to achieve adaptive sampling of a given oceanographic area. Different from Leonard et al. (2007), in Caiti et al. (2007) the metric for adaptive sampling is based on a deterministic estimate of the oceanographic field, through the use of radial basis functions interpolators, instead of a stochastic metric. A cooperative algorithm does not enforce any particular formation to be maintained by the AUV team, but it does enforce a set of mission-related optimality conditions to be collectively satisfied by the team.

This paper is composed of three parts: in the first part, the Fòlaga III vehicle, the last evolution of the class, is described, with discussion of its hardware and software components, payloads, mission management and current developments; the second part discusses the vehicle navigation and control system and presents examples of vehicle behaviour from experimental data; in the third part, a novel cooperative algorithm for adaptive oceanographic sampling is proposed. While the work on AUV cooperation is motivated by the availability of the Fòlaga vehicles, it is explicitly remarked that the cooperative algorithm is for general use, and not restricted by the vehicle characteristics.

The first part in this paper is mostly descriptive: it provides a uniform and coherent presentation of material appeared before in conference presentations, plus some original material, in particular as for the vehicle diving: in the Folaga III development, here reported, diving is achieved through a combination of buoyancy change (as in glider navigation, Jones et al., 2005) and attitude change through the movement of the internal vehicle mass; the Folaga III does not rely on hydrodynamic surfaces and it does not need a minimum speed for manoeuvring. The second part discusses the rationale for the implementation of controllers by combining a (possibly) nonlinear kinematic control with robust PD control, following a back-stepping methodology proposed originally by Qu and Dorsey (1991) and developed or applied in numerous other papers thereafter, as for instance, to limit ourselves to methodological approaches in mechanical systems control, (Berghuis and Nijmeijer, 1994; Canudas de Wit et al., 1994). The methodology guarantees the uniform ultimate boundedness (i.e. the practical stability) of systems under PID control providing that the uncertainties and the nonlinearities of the system can be bounded by a polynomial function of the error norm. The surface navigation kinematic controller is a straightforward application of the unicycle control of Aicardi et al. (1995), however, with addition of an integral term to counteract sea current effects; the resulting discussion on the implications for surface navigation, including simulative examples, has not appeared before, although the same approach has been adopted in Caccia and Veruggio (2000). The discussion on the diving phase control is novel and motivated by the Fòlaga III peculiarities: the actuation mechanism for diving introduces an additional difficulty in terms of a thrust allocation problem, that can still be tackled in practical terms with PID design, but with additional care in the controller tuning. Previously unpublished experimental data are reported to show that this has been accomplished for the Fòlaga III case. The third part of the paper is entirely original: it builds up on the adaptive oceanographic sampling setting of Caiti et al. (2007), adding underwater communication constraints, in terms of range limitation between pairs of vehicles. By imposing a serial hierarchical structure on the vehicle team, the constrained adaptive oceanographic sampling problem has a solution in terms of dynamic programming, which in turn can be implemented in a distributed fashion. In addition to the nice mathematical structure of the distributed dynamic programming, the simulative application of the algorithm shows that the AUV team moves in formation, although vehicle formation was not considered in the problem setting, sometime reconfiguring the formation accordingly to the current sampled field information and mission status. The emerging coordinated behaviour of cooperating vehicles in this context is considered a relevant and intriguing result, even beyond the domain of AUVs and oceanic engineering applications.

Section snippets

Fòlaga III: design, components and system characteristics

The Fòlaga project is a combined research effort in which the IMEDEA Institute has provided the oceanographic specifications and requirements, ISME has provided the system design, GraalTech (a spin-off company of the University of Genova, Italy) has taken care of the realization (mechanics, power, electronics, software implementation) and NURC has provided advice for hydrodynamics, sealing, wet/dry part connections and, in the most recent developments, acoustic communication. Presently, there

Navigation and control

Navigation and control choices are now described, referring in particular to the vehicle navigation at the sea surface and to the diving phase. In both cases, the following approach has been followed: starting from the kinematic equations of a possibly simplified model, a kinematic controller has been derived, i.e., a controller using the velocities in the kinematic model as control actions. The velocities so derived are used as reference velocities in a velocity controller, which in turn is

Towards AUV team cooperation: an on-line planning algorithm for adaptive oceanographic sampling

In oceanographic sampling the goal is to produce a map of a given environmental quantity (e.g., temperature, or sound speed, or bottom morphology, depending on the specific payload) accurately and in the minimum amount of time. When 100% coverage is not strictly required, or it is not possible because the payload can only make a point-wise measurement (as in the temperature/salinity case), the produced map is an estimate of the true map based on the available samples. To meet the accuracy

Conclusions

The development and design choices of a low-cost, light-weight AUV, the Fòlaga III, for coastal oceanography mission have been described. The vehicle has a diving actuation mechanism as an oceanographic glider and autonomous surface navigation capabilities as self-propelled AUVs. Surface navigation relies on the approach developed for unicycle-like vehicles. Control of both surface and diving phases is implemented with a two-step approach, based on the theory of robust PID and back-stepping.

Acknowledgments

This work was partially supported by the European Union and Regione Liguria, under the PRAI-FESR initiative, by the Italian Ministry of Research, Projects of National Interest, Project PICTURE, and by Spanish and Italian Ministries of State, under the joint Italy–Spain action for research developments and exchange.

References (24)

  • M. Caccia et al.

    Guidance and control of a reconfigurable unmanned underwater vehicle

    Control Engineering Practice

    (2000)
  • M. Porfiri et al.

    Tracking and formation control of multiple autonomous agents: a two-level consensus approach

    Automatica

    (2007)
  • M. Aicardi et al.

    Closed-loop steering of unicycle-like vehicles via Lyapunov techniques

    IEEE Robotics and Automation Magazine

    (1995)
  • Aicardi, M., Casalino, G., Indiveri, G., 2001. Closed loop time invariant control of 3D underactuated underwater...
  • Alvarez, A., Caffaz, A., Caiti, A., Casalino, G., Clerici, E., Giorgi, F., Gualdesi, L., Turetta, A., 2004. Design and...
  • Alvarez, A., Caffaz, A., Caiti, A., Casalino, G., Clerici, E., Giorgi, F., Gualdesi, L., Turetta, A., Viviani, R.,...
  • Araki, S., Ishi, K., 2007. Development of glider type small AUV Seabird. In: Proceedings of the IEEE International...
  • H. Berghuis et al.

    Robust control of robots via linear estimated state feedback

    IEEE Transactions on Automatic Control

    (1994)
  • Bozzo, T., Caffaz, A., Clerici, E., Giorgi, F., Caiti, A., Casalino, G., Turetta, A., Viviani, R., Gualdesi, L., 2005....
  • A. Caiti et al.

    Adaptive on-line planning of environmental sampling missions with a team of cooperating autonomous underwater vehicles

    International Journal of Control

    (2007)
  • Canudas de Wit, C., Berghuis, H., Nijmeijer, H., 1994. Practical stabilization of nonlinear systems in chained form....
  • El. Desa et al.

    A small autonomous surface vehicle for ocean color remote sensing

    IEEE Journal of Oceanic Engineering

    (2007)
  • Cited by (175)

    • Review of low cost micro remotely operated underwater vehicle

      2022, Ocean Engineering
      Citation Excerpt :

      The closed-loop control system has been known to improve the efficiency of ROV devices. Numerous researches regarding the implementation of PID controllers are found in (Hoang and Edwin Kreuzer, 2007; Alvarez et al., 2009; Özgür et al., 2009; Azmi et al., 2018; Walker et al., 2020). Sliding mode controller (SMC):- The SMC is a nonlinear system that employs a discontinuous control system, forcing the system to slide along the cross-section of the predicted normal behavior, through which it alters the dynamics of the nonlinear system.

    • Hydrodynamic design and analysis of a hybrid-driven underwater vehicle with ultra-wide speed range

      2022, Ocean Engineering
      Citation Excerpt :

      It is characterized by the bionic hull, wing or fin structures and corresponding bionic actions, such as the dolphin-like glider (Wu et al., 2016), the twin flapping foils (Li and Ma, 2020), the bionic batfish glider (Wang and Zhang, 2021), the webbed wings (Tian and Zhou, 2016) and the pectoral fin and caudal actuation (Fu et al., 2019). Third, by assembling the additional propulsion devices at the tail or around, the AUG can obtain various sailing modes driven by buoyancy, thrust or both, such as the Sterne (Graver, 2005), Folaga (Alvarez et al., 2009) and Petrel (Wang et al., 2010). This kind of AUG can cruise at about 3 kn, but it will be limited by the large resistance of the wings at higher speed.

    View all citing articles on Scopus
    View full text