Side Scan Sonar-based SLAM for Autonomous Algae Farm Monitoring

TL;DR

The paper tackles safe autonomous navigation for AUVs in kelp farms by introducing a side scan sonar–based SLAM framework that exploits the farm's rope geometry. It treats each rope detection as an independent landmark with shared rope priors, creating a sparse factor graph optimized online with iSAM2, which yields robust online pose and map estimates despite drift. Compared to two baselines, the method achieves lower mapping errors (rope RMSE ~1.00 m, buoy RMSE ~1.14 m) and smoother trajectory estimates, while still operating in real time on onboard hardware. This approach paves the way for reliable, scalable autonomous monitoring of marine farming infrastructure, reducing the need for surface vessels and divers.

Abstract

The transition of seaweed farming to an alternative food source on an industrial scale relies on automating its processes through smart farming, equivalent to land agriculture. Key to this process are autonomous underwater vehicles (AUVs) via their capacity to automate crop and structural inspections. However, the current bottleneck for their deployment is ensuring safe navigation within farms, which requires an accurate, online estimate of the AUV pose and map of the infrastructure. To enable this, we propose an efficient side scan sonar-based (SSS) simultaneous localization and mapping (SLAM) framework that exploits the geometry of kelp farms via modeling structural ropes in the back-end as sequences of individual landmarks from each SSS ping detection, instead of combining detections into elongated representations. Our method outperforms state of the art solutions in hardware in the loop (HIL) experiments on a real AUV survey in a kelp farm. The framework and dataset can be found at https://github.com/julRusVal/sss_farm_slam.

Figures (7)

• Figure 1: Aerial view of the algae farm (top) with a depiction of the survey plan and the AUV (bottom) used for the survey.
• Figure 2: Diagram of an algae farm infrastructure. The algae hangs from the horizontal lines supported by the buoys (mooring buoys in yellow, intermediate buoys in white). The thicker lines and the gray boxes indicate mooring lines and anchors, respectively.
• Figure 3: Depiction of two parallel SSS swaths from combining consecutive SSS pings (purple and green) collected along a farm line (yellow). 2D rope detections in the SSS imagery (green and purple circles) are used to extract the rope coordinates through the SSS geometry model, in red. The misalignment in the detections is due to the DR drift.
• Figure 4: Factor graph formulations of the presented approach (top) and the baseline method from fallon2011efficient (bottom). Variable nodes are indicated by white circles: $m_t$ represents the AUV states, $b_j$ represents the buoy positions, and $r_l$ represents the rope positions. Factor nodes are indicated by colored circles: yellow for priors on the initial state of the AUV, magenta for priors on buoys, blue for priors on ropes, red for odometry measurements, and green for detection measurements.
• Figure 5: SSS detector results referenced to the aerial image of the algae farm. The numerals correspond to the pass number of the AUV survey, in Fig. \ref{['fig:farm_real']}. Red markers show buoy and green, individual rope detections. Note the missed buoy on IV.