Boundary Control Behaviors of Multiple Low-cost AUVs Using Acoustic Communication

TL;DR

This work tackles boundary control for multiple low-cost AUVs using acoustic communication by proposing two models: Range Variation-Based (RVB) and Heading Estimation-Based (HEB). RVB relies solely on range data to enforce Fencing and Milling around a central beacon, while HEB uses range rates to estimate the beacon direction $\theta$ and enables more flexible boundary and path shapes. The methods are evaluated through simulations, pool experiments with currents, and field tests in shallow coastal waters, showing trade-offs: HEB generally achieves tighter path adherence on simple shapes, while RVB demonstrates robustness with limited information. The results support scalable, autonomous multi-AUV coordination using a single beacon, with practical implications for underwater swarm robotics and environmental sensing.

Abstract

This study presents acoustic-based methods for the control of multiple autonomous underwater vehicles (AUV). This study proposes two different models for implementing boundary and path control on low-cost AUVs using acoustic communication and a single central acoustic beacon. Two methods are presented: the Range Variation-Based (RVB) model completely relies on range data obtained by acoustic modems, whereas the Heading Estimation-Based (HEB) model uses ranges and range rates to estimate the position of the central boundary beacon and perform assigned behaviors. The models are tested on two boundary control behaviors: Fencing and Milling. Fencing behavior ensures AUVs return within predefined boundaries, while Milling enables the AUVs to move cyclically on a predefined path around the beacon. Models are validated by successfully performing the boundary control behaviors in simulations, pool tests, including artificial underwater currents, and field tests conducted in the ocean. All tests were performed with fully autonomous platforms, and no external input or sensor was provided to the AUVs during validation. Quantitative and qualitative analyses are presented in the study, focusing on the effect and application of a multi-robot system.

Figures (20)

• Figure 1: Overview of behaviors. (a) Fencing: the AUV is free to move within the boundary, but when it exceeds the boundary, the algorithm guides the AUV back towards the center. (b) Milling: The algorithms guide the AUV on a closed loop path around the center.
• Figure 2: RVB Fencing ranges representation. Shown AUV position at times $t_i,t_{i-1},t_{i-2}$. As the AUV leaves the boundary, the last three ranges are recorded ($r(t_{i}),r(t_{i-1}),r(t_{i-2})$), and the range variation calculated ($\Delta r_i, \Delta r_{i-1}$).
• Figure 3: Possible conditions of AUV during Milling. Condition (a) represents the scenario where the AUV is beyond the boundary but approaching it. In condition (b), AUV is within the boundary and moving towards the center. Condition (c) depicts the AUV inside the boundary and moving away from the center. Condition (d) occurs when the AUV is outside the boundary and moving away.
• Figure 4: HEB model representation for $\theta$ estimation
• Figure 5: HEB Fencing on 3 boundaries for 3 AUVs. (a) Circle, (b) Square, (c) Isotoxal Star