Master thesis proposals

Simultaneous Localization and Mapping with autonomous underwater vehicles

Start : ASAP
Contact : Ignacio Torroba Balmori (torroba@kth.se)
Keywords : AUV, SLAM, sonar.

The growing need to better understand and protect our oceans worldwide motivates the research within the Swedish Maritime Robotics Center (SMaRC) on the use of autonomous underwater vehicles (AUVs) as a tool towards this end. Among our target scenarios are the use of AUVs for exploration of our largely unknown oceans, climate change study and glaciology or algae farming.

A basic capability of any autonomous mobile robot is the ability to build maps of its surroundings and localize itself with respect to them in order to navigate safely. This is particularly crucial for AUVs on large, underwater areas, where errors in the pose estimate might lead to the vehicle being completely lost. The problem of building a map of an unknown terrain while localizing wrt it is referred to as SLAM, and it is currently being researched intensively across domains and platforms. However, the underwater environment poses particular challenges such as lack of GPS, reduced sensing and communication capabilities or large and dynamic environments, which make UW SLAM an open problem yet.

On this project we are aiming at implementing a SLAM framework based on multibeam (MBES) and sidescan (SSS) sonar measurements in one of the SMaRC AUVs, Lolo (see picture below). The work will be based on current research from this group in bathymetric Graph SLAM and loop closure identification through sonar data [1, 2]. The student will work within the SMaRC team, building upon and integrating our previous work on a real robot and using real data collected with several vehicles from industry and university partners. This could include participating in data collection and/or sea trials of the final system. Candidates are expected to have strong C++/python programming skills, knowledge of ROS and a background in robotics and applied estimation.

 

[1] Torroba, Ignacio, et al. "PointNetKL: Deep Inference for GICP Covariance Estimation in Bathymetric SLAM." IEEE Robotics and Automation Letters 5.3 (2020): 4078-4085.

[2] Torroba, Ignacio, Nils Bore, and John Folkesson. "Towards autonomous industrial-scale bathymetric surveying." 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2019, 3-8 November 2019, Macau, China. Institute of Electrical and Electronics Engineers Inc., 2019.