Skip to main content
To KTH's start page To KTH's start page

Hydrobatics: Real-time Control, Simulation and Learning for Underactuated AUVs in Agile Maneuvers

Time: Fri 2023-10-06 14.00

Location: Kollegiesalen, Brinellvägen 8, Stockholm

Video link:

Language: English

Subject area: Vehicle and Maritime Engineering

Doctoral student: Sriharsha Bhat , Farkostteknik och Solidmekanik

Opponent: Massimo Caccia, Consiglio Nazionale delle Ricerche (CNR)

Supervisor: Ivan Stenius, Farkostteknik och Solidmekanik; Dimos V. Dimarogonas, Reglerteknik, Centrum för autonoma system, CAS, ACCESS Linnaeus Centre

Export to calendar


The term hydrobatics refers to the agile maneuvering of underwater vehicles. Underwater robots such as autonomous underwater vehicles (AUVs) and remotely operated vehicles (ROVs) are either designed as flight style, optimized for range and speed, or hover style, optimized for precise maneuverability. Hydrobatic capabilities can help balance efficiency and maneuverability in these platforms, enabling innovative robot designs for impact areas in environmental monitoring, ocean production and security. This dissertation addresses technical challenges related to hydrobatic AUVs and contributes to new knowledge in real-time control, simulation, learning and planning. 

Hydrobatic AUVs are underactuated systems --- new strategies using nonlinear model predictive control (MPC) and behavior trees (BTs) are presented for efficient and safe real-time control of underactuated AUVs in agile maneuvers. Further, the flow around an AUV during such maneuvers transitions from laminar to turbulent flow at high angles of attack, rendering flight dynamics modelling difficult. A full 0-360 degree envelope flight dynamics model is therefore derived, which combines a multi-fidelity hydrodynamic database with a generalized component-buildup approach. Such a model enables real-time (or near real-time) simulations of hydrobatic maneuvers including loops, helices and tight turns. To increase the intelligence and robustness of such systems, data driven methods including physics-informed learning, Gaussian processes, sparse regression  and reinforcement learning are utilized to rapidly identify models of the system's dynamics and perform online adaptive control. To further enhance autonomy, informative path planning is also studied, where an adaptive sampling strategy combines AUV measurements and satellite data to track ocean fronts.

These hydrobatic capabilities are safely brought to the real world through a cyber-physical system (CPS). Simulator environments are closely integrated with the robotic system, enabling pre-validation of controllers and software before hardware deployment. The small and hydrobatic SAM AUV (SAM: Small and Affordable Maritime robot) developed in-house at KTH as part of the Swedish Maritime Robotics Centre (SMaRC) is used as a test platform. The CPS concept is demonstrated with the SAM AUV in applications including detecting underwater targets, inspecting seaweed farm infrastructure and tracking algal blooms using the presented simulation, planning and control strategies.