Towards Autonomous Industrial-Scale Bathymetric Surveying

Abstract

Both higher efficiency and cost reduction can be gained from automating bathymetric surveying for offshore applications such as pipeline, telecommunication or power cables installation and inspection on the seabed. We present a SLAM system that optimizes the geo-referencing of bathymetry surveys by fusing the dead-reckoning sensor data from the surveying vehicle with constraints from the maximization of the geometric consistency of overlapping regions of the survey.The framework has been extensively tested on bathymetric maps from both simulation and several actual industrial surveys and has proved robustness over different types of terrain. We demonstrate that our system is able to maximize the consistency of the final map even when there are large sections of the survey with reduced topographic variation. The framework has been made publicly available together with the simulation environment used to test it and some of the datasets.

PointNetKL: Deep Inference for GICP Covariance Estimation in Bathymetric SLAM

“…In the remainder of this paper we apply the presented general approach for learning Gaussian distributions from unordered sets of points to the specific problem of bathymetric graph SLAM with GICP registration as introduced in [21]. The reason to focus on GICP as opposed to ICP is that this method works better on the kind of bathymetric point clouds produced from surveys of unstructured seabed, as discussed in [6].…”

“…Q ij becomes Q i (which is also approximately Q j ). The validity of this is based on the fact that it is possible to aggregate raw sonar data on fairly uniform point clouds with the submaps approach used in [21], for example.…”

“…For the tests, two underwater surveys outside the training dataset of the network have been used, named Bornö 8 and Thwaites 11. For the test on each scenario, a bathymetric pose graph is created and optimised similarly to [21]. The graph optimisations have been run with three different sets of covariances: those obtained with our method, the ones approximated with MC method and a constant covariance for each experiment.…”

“…The aim of these tests is to assess the influence of the GICP covariances on the quality of the PoseSLAM solution. To this end, two modifications have been introduced in the construction of the pose graph with respect to [21]:…”

“…Algorithm 1 outlines the process followed to create the graphs for the optimisation tests. As input it requires a 'GT' dataset, which we approximate optimising the navigation estimate from the vehicle DR as in [21] using the MC covariances computed. Given the estimated GT dataset divided into a set of N submaps S N S N S N , a graph G G G is constructed in lines 2 to 12.…”