Localization and control of a string of underwater robots in confined environments

Juliette DRUPT (Oct. 2020)

Financing: ministerial grant


  • Vincent HUGEL, University Professor (director)
  • Claire DUNE, Assistant Professor at the University of Toulon (co-supervision),
  • Andrew Comport, CNRS I3S research fellow (co-supervision).


If autonomous submarines for mapping in open water are now operational, the exploration of confined environments remains a challenge in underwater robotics: cenotes, caves, wrecks, mines,…  The exploration of submerged mines has been the subject of recent H2020 projects whose objective was to evaluate the state of the structures in order to prevent their collapse and to estimate the quantity of ore stocks, but none of these projects was interested in the management of the cable that links the robot to the surface.

Figure 1: 1. In underwater robotics, electromagnetic waves are absorbed within the first few meters, isolating the robots from the surface. The only way to maintain a high speed communication in real time is to use an umbilical. 2. This cable, deployed over a large distance, can disturb the control of the robot. 3. We propose to add intermediate robots regularly on the cable to control its shape, it is the concept of ROBOTS CORD.

The exploration of confined underwater environments requires long diving times and a permanent adaptation of the movements according to the discovered environment. Only remotely operated vehicles (ROVs) combine these two skills. Indeed, divers have relatively short diving times and can only access limited depths. When it comes to autonomous robots, without umbilicals, of the AUV (autonomous underwater vehicle) type, despite advances in artificial intelligence and energy storage, they are still very limited in terms of diving time and decision making. ROVs benefit from an unlimited energy supply and the support of a remote operator who co-analyzes the data from the sensory envelope transmitted in real time by the cable. 

Far from being only a disadvantage, the cable offers a mechanical support to recover the robot and can be used as a breadcrumb trail to find the exit of the explored network.  However, to deploy a cable system in a confined environment, a method must be developed to control the position of the cable to prevent it from becoming entangled or jammed. Automatic management of the cable is a problem that has not been studied, or has been studied very little, until now (for ROV operation, a second operator supervises the deployment of the cable). 

We propose to control the cable by adding reeling systems and intermediate mini-robots evenly distributed along the cable run. We have named this concept ROBOTS CORD. 

This project has become one of the research axes of the Cosmer laboratory and is already the support of two theses. In a first thesis [Laranjeira2017,Laranjeira2018, Laranjeira2019, Laranjeira2020], we demonstrated that it is possible to control a pair of robots linked by a weighted cable using the image of this cable filmed by the cameras embedded in the system. A second ongoing thesis focuses on the physical modeling of a cable and aims at developing a dynamic tension control system via the design of an active reel [Tortorici 2020].

Figure 2: Control of an underwater robot (main view) from the acquisition of images of its mooring line (top left view). 

The subject of the thesis that we propose here is an extension of the thesis of Matheus Laranjeira. The objective is to control a rope of robots in order to explore and map confined submerged places.

  1. SLAM: Environment Mapping and Simultaneous Position Estimation of an Underwater Robot: the first work consists in evaluating and adapting mapping and simultaneous localization algorithms for marine environments [Vidal2017, Weidner2017, Meilland2013, Mahe2019, Rahman2019]. Terrestrial localization techniques can be applied to the underwater environment but require adaptations. Indeed, they are developed for structured and static environments, such as the streets of a city or the interior of a building. In the underwater environment, a large part of the scene is in motion and, in the case of natural environment, the salient points are less numerous. On the other hand, the image formation model differs in water: absorption and backscattering parameters are added to the usual model of light wave propagation in air. This part will be based on an adaptation of Andrew Comport’s work for aerial cameras.
  2.  T-SLAM: Simultaneous map co-construction and rope parameter estimation: the second step is to simultaneously deploy the underwater SLAM on all modules regularly distributed on the Robot Rope. The sensors (cameras, inertial units, depth gauges) are distributed along the rope and thus constrained by its shape. They allow to co-construct a map of the environment and to estimate the parameters of a model of the complete system. The constraints imposed by the shape of the cable and its length can be merged with the information acquired by the sensors to refine the map obtained: T-SLAM (Tethered Simultaneous Localisation and Mapping) [McGarey2017]. This second part will be based on results obtained at the COSMER laboratory on the modeling of an umbilical [Laranjeira2020, Tortorici 2020].
  3. Trajectory planning and control of a robot rope in a confined environment: in the rope, if the leading robot is controlled by an operator, on the other hand, the movements of the intermediate robots and the cable must be automated. It is then necessary to control a mixed cable-robot dynamic system including a dynamic model of the cable and the management of the platooning effects (accordion effects due to the delays of the reactions of each module, equivalent to the effects observed in the trains of vehicles). This part will be based on the first results of the control of a two-vehicle rope obtained during the thesis of Matheus Laranjeira [Laranjeira2020] and on the mappings performed in points 1 and 2.

Figure 3: Two BlueRov mini underwater robots roped in the wave basin of the University of Toulon [Laranjeira2020].



[Laranjeira2017] Matheus Laranjeira, Claire Dune, Vincent Hugel. Catenary-based visual servoing for tethered robots IEEE International Conference on Robotics and Automation (ICRA), May 2017, Singapour, France 

[Laranjeira2018] Matheus Laranjeira, Claire Dune, Vincent Hugel, Local Vision-Based Tether Control for a Line of Underwater Robots, Conference IEEE IROS 2018 Madrid, Workshop on New Horizons for Underwater Intervention Missions: from Current Technologies to Future Applications.

[Laranjeira2019] Matheus Laranjeira, Claire Dune, Vincent Hugel. Embedded Visual Detection and Shape Identification of Underwater Umbilical for Vehicle Positioning OCEANS 2019 – Marseille, Jun 2019, Marseille, France. Pp.1-9

[Laranjeira2020] Matheus LaranjeiraClaire DuneVincent HugelCatenary-based visual servoing for tether shape control between underwater vehicles Ocean Engineering, Elsevier, 2020, 200, pp.107018 ⟨10.1016/j.oceaneng.2020.107018⟩

[Mahe2019] Real-time RGB-D semantic keyframe SLAM based on image segmentation learning from industrial CAD models, Howard Mahe, Denis Marraud, Andrew I. Comport,  International Conference on Advanced Robotics, Dec 2019, Belo Horizonte, Brazil.

[McGarey2017] McGarey, P., MacTavish, K., Pomerleau, F., & Barfoot, T. D. (2017). TSLAM: Tethered simultaneous localization and mapping for mobile robotsThe International Journal of Robotics Research36(12), 1363–1386. https://doi.org/10.1177/0278364917732639

[Meiland2013] 3D High Dynamic Range Dense Visual SLAM and Its Application to Real-time Object Re-lighting. Maxime Meilland, Christian Barat, Andrew I. Comport. International Symposium on Mixed and Augmented Reality, Oct 2013, Adelaide, Australia

[Rahman2019] Rahman, Sharmin & Quattrini Li, Alberto & Rekleitis, Ioannis. (2019). SVIn2: An Underwater SLAM System using Sonar, Visual, Inertial, and Depth Sensor. 1861-1868. 10.1109/IROS40897.2019.8967703. 

[Tortorici 2020] Ornella Tortorici, Cédric Anthierens, Vincent Hugel, Herve Barthelemy. Towards active self-management of umbilical linking ROV and USV for safer submarine missionsIFAC-PapersOnLine, Elsevier, 2019, 52 (21), pp.265 – 270. ⟨10.1016/j.ifacol.2019.12.318⟩⟨hal-02428777⟩

[Vidal2017] Eduard Vidal, Juan David Hernandez, Klemen Istenic and Marc Carreras, Online View Planning for Inspecting Unexplored Underwater StructuresIEEE Robotics and Automation Letters2, 3, (1436), (2017).

[Weidner2017] N. Weidner, S. Rahman, A. Q. Li and I. Rekleitis, « Underwater cave mapping using stereo vision, » 2017 IEEE International Conference on Robotics and Automation (ICRA), Singapore, 2017, pp. 5709-5715. doi: 10.1109/ICRA.2017.7989672