Back to search

FRINATEK-Fri prosj.st. mat.,naturv.,tek

An underwater robotics concept for dynamically changing environments

Alternative title: Undervannsrobotikk i dynamiske omgivelser

Awarded: NOK 10.1 mill.

The CHANGE project is led by the young research talent Dr. Eleni Kelasidi at SINTEF Ocean, who will work in cooperation with the NTNU and MIT. Together, they will generate fundamental knowledge on modelling of UUVs and develop advanced control strategies for UUVs interacting with complex environments. The developed models will include fish behavior and deformation of flexible structures, combined with influences from the surrounding environment (e.g. currents, waves), thus providing data for feedback control loops. Integrating these with the novel control strategies will enable real-time control of the UUV during autonomous navigation in aquaculture fish cages without colliding with fish or flexible structures despite variable currents or waves. By enabling UUVs to adapt their actions to the dynamically changing environment, CHANGE will promote operational efficiency, safety, and the sustainable expansion of Norwegian salmon farming. Several operations in fish farms involve adaptation and use of technological tools. To understand and optimize the use of tools in fish farms we obtained studies to evaluate the behavior change of fish. Therefore, in 2022, several experiments have been conducted to study fish behavior change during different impact factors in industrial scale fish farms. The gathered data have been used to develop methods to identify and quantify the change on behavior of fish under the impact of different conditions during autonomous operations in fish farms. Identifying which parameters can trigger fish behavior change, we are able to derive rules and requirements on which type of technological solutions should be adapted for use in fish farms and develop in the future solutions that affect minimum the fish. The obtained results will also be used for adaptation of existing fish behavior models and will provide essential inputs on the new methods to be adapted and developed for control of used underwater vehicles in fish farms. In 2022 the project also targeted the development of integrated system incorporating models of fish behavior, deformable structures and effects from the surrounding environment which can be utilized to provide inputs for feedback control loops. Furthermore, the models of flexible structures have been validated in fish farms and autonomous navigation of remotely operated vehicle is demonstrate at SINTEF ACE for high quality data acquisition. Integrating these models with the novel control strategies will enable real-time control of the UUV during autonomous navigation in aquaculture fish cages without colliding with fish or flexible structures despite variable currents or waves. The team has also targeted the implementation of models of remotely operated vehicles and investigated motion planning methods and control strategies for obstacle avoidance during operations in fish farms. The functionality of the general framework consisting of models and control strategies for obstacle avoidance have been tested in laboratory experiments and field tests at SINTEF ACE facility. In 2023, the project performed more extensive field trials to gather data to study the fish behaviour change under the influence of different impact factors. In particular, we have studied the effects of sound, light, shape, size and color of structures on fish behaviour change. Neural Network (NN) methods have been developed to identify the fish behaviour change utilizing both acoustic and vision data. In addition, NN methods have been evaluated to identify both local and global fish behaviour integrating simulated and experimental data. In addition to the NN methods, methods based on statistics will be utilized to validate the current outcomes. In March 2023, a second round of experiments have been conducted to collect data and study Robot-Fish Interaction. Several results have been obtained when it comes to modeling of structures, underwater vehicles (remotely operated vehicles and underwater snake robots) and control design. The project developed digital twin concepts for aquaculture integrating models of the environment, structures, fish, robotic systems, estimation techniques, navigation and control concepts. Novel guidance and several motion planning methods have been developed suited for underwater vehicles operating in highly dynamic environments such the ones faced in aquaculture. All methods aimed at demonstrating zero-collision capabilities during autonomous operations of underwater vehicles. Field trials have been performed in industrial scale fish farms (SINTEF ACE facilities) to demonstrate autonomous inspection operations while avoiding static and moving obstacles inside the net pen. In 2023, the project also targeted research on robust localization. Several sensors have been integrated on the robotic systems to gather data and compared the efficacy of different localization methods in fish farms.

As salmon farm sites are moved further offshore and to more exposed locations, working conditions are increasingly challenging. Farmers therefore aim to automate certain operations to facilitate safer working conditions. Automation and autonomous unmanned underwater vehicles (UUVs) are furthermore key elements in meeting the desire for increased precision in finfish farming that will enable aquaculture to advance operational efficiency, safety and thus sustainability. While current models and control strategies for UUVs allow navigation among rigid structures in static environments, they are not sufficient for UUV operations in a dynamic fish farm environment where the UUV needs to react to the presence of animals and deformable structures influenced by external forces such as waves and currents. In the CHANGE project, fundamental knowledge on modelling of UUVs interacting with complex environments will be developed together with advanced control strategies. The developed models will include fish behaviour and deformation of flexible structures, combined with influences from the surrounding environment (e.g. currents, waves), thus providing data for feedback control loops. Integrating these with the novel control strategies, will enable real-time control of the UUV during autonomous navigation in aquaculture fish cages without colliding with fish or flexible structures despite variable currents or waves. The functionality of the resulting new control paradigm will be tested in laboratory experiments that mimic the dynamic environments of aquaculture sites. Final field tests at active salmon farms will be employed to validate the developed models and strategies during demanding operations. By enabling UUVs to adapt their actions to the dynamically changing environment, CHANGE will promote the sustainable expansion of Norwegian salmon farming while simultaneously offering opportunities also for application outside the aquaculture context.

Publications from Cristin

Funding scheme:

FRINATEK-Fri prosj.st. mat.,naturv.,tek