Prev-Talk | Next-Talk | All-Talks | Talks-Sorted

Talk-21: Cooperative Seabed Coverage with MOOS-IvP, LCM, iSAM, Goby, and HoverAcomms

Liam Paull, MIT

Mae L. Seto, Defence Research and Development Canada

John J. Leonard, MIT

Many typical applications in marine robotics such as seabed imaging and data sampling fall under the broad umbrella of robotics area coverage tasks. In the absence of topside assets or pre-installed beacon infrastructure, a submerged AUVs position uncertainty will grow without bound. Underwater area coverage tasks should explicitly account for the platform uncertainty to quantify coverage performance.

Area coverage is also a task that is inherently parallelizable. The area to be covered can often be decomposed into N sub-areas which are allocated to each of a team of N robots, resulting in a factor of N increase in the rate of coverage. In the case that the robots can also communicate and observe each other, then the advantage of the team-based approach is even greater since the vehicles can use these measurements and communications to reduce their respective rates of uncertainty growth. This is the case for marine vehicles communicating through acoustics who possess highly synchronized onboard clocks. However, communication through acoustics has a number of challenges and requires special packet generation schemes to ensure robustness.

The evolution of uncertainty in the system and the ability to communicate is stochastic in nature and therefore preplanning paths is insufficient. Instead, an adaptive coverage planner that is able to exploit the measurements available to optimize the rate of coverage given the sensor uncertainty is developped. In short, the planner automatically spaces tracks more sparsely when localization accuracy is better and vice-versa.

Since this work requires realtime acoustic packet generation and planning, validation by post-processing of data is not possible. Instead, a realistic simulation environment has been built that combines MOOS-IvP for 2D visualization, vehicle simulation, autonomy, and control, lightweight communication and marshaling (LCM) and incremental smoothing and mapping (iSAM) for pose graph optimization and packet generation, the Goby suite of tools for acoustic packet encoding and decoding, and the HoverAcomms library for acoustic simulation and processing. All of these components work in concert to allow rapid transition onto real robotic hardware. In this talk, we will describe the software framework and show how it is applied to robust and cooperative realtime coverage planning with a particular focus on the seabed mapping for mine countermeasures problem. DOWNLOAD


  • UUVs
  • Navigation/SLAM
  • Seafloor Surveying
  • Sensors/Sonar