Efficient Control Allocation and 3D Trajectory Tracking of a Highly Manoeuvrable Under-actuated Bio-inspired AUV
This paper presents a novel control allocation strategy for underactuated, bio-inspired autonomous underwater vehicles (AUVs) driven by fin actuators. The primary focus is on achieving high maneuverability, energy efficiency, and computational efficiency in tracking complex 3D trajectories. The researchers introduce an adaptive hybrid feedback controller that enables an AUV with only four actuators to maneuver in six degrees of freedom (DOF) in simulation environments and up to five DOF in practical pool experiments. The proposed methodology stands out with centimeter-scale tracking accuracy, outperforming existing state-of-the-art control allocation techniques.
The fin actuators are pivotal in this paper as they facilitate thrust generation and vectoring with fewer components than traditional propeller-based systems. However, controlling such fin-driven vehicles poses significant challenges due to the coupling between thrust direction and magnitude. These complexities are managed by an innovative control framework that balances between navigational prowess and energy consumption.
In the simulation tests, the AUV's fin-controlled movements were analyzed using multiple trajectory profiles, such as ellipsoid and Lissajous patterns. The proposed control allocation method demonstrated a marked improvement over conventional techniques in both operational performance and resource consumption. Metrics such as root mean squared error (RMSE) show that the AUV maintained high accuracy in trajectory tracking with minimal actuation force demands. Additionally, the computational load of control allocation was reduced significantly, enhancing real-time performance potential on embedded systems like the Jetson TX2 processor.
Further supporting this, pool-based trials confirmed that the controller could robustly and precisely follow desired paths within operational constraints, offering greater control fidelity compared to extant approaches.
The implications of this work underscore a substantial leap in the applicability of bio-inspired AUVs. By leveraging fewer actuators, this method paves the way for constructing more compact and efficient underwater vehicles appropriate for practical applications such as marine biology studies, underwater archaeology, and environmental monitoring. The reductions in energy and computation cost are particularly promising for prolonged autonomous missions where resource constraints are critical considerations.
Future research directions could extend this control paradigm to address fault-tolerant capabilities, extended thrust models to simulate complex fluid dynamics more accurately, and real-world validations beyond static pool environments. Potential adaptations of reinforcement learning could further refine control algorithms dynamically, providing more nuanced responses to environmental variability. Overall, this paper contributes an important framework that situates bio-inspired AUVs as viable, versatile platforms in underwater robotics.