River Mapping via Autonomous Surface Vehicles (ASVs)

River flooding is among one of the most hazardous natural disasters in the world. Over the past ten years, river floods across the globe resulted in a staggering average of 50 billion dollars a year. Most rivers are conduits of water and sediments and their geometry evolves over time due to erosion and deposition. It is possible to determine where and when a flooding might occur given an accurate profile of river geometry.

The goal of this project is to apply the techniques of autonomous navigation and state estimation in river mapping. We are configuring an Autonomous Surface Vehicle (ASV) and equipping it with an Acoustic Doppler Current Profiler (ADCP) to create high resolution river bed topographies and flow rate profiles. Although autonomous vehicles have been used extensively to create topography of oceans and lakes, much less work has been done for river beds because of the difficulties associated with turbid current and shallow depths. Although lawnmower pattern paths may provide adequate coverage for most still water environments, high current flow in rivers might make certain regions impossible to traverse. This project will aim to apply real time path planning techniques such as A*, RRT*, or other path planning algorithms to navigate in turbid river environments. After each mission, we will fuse data collected during the deployment to create spatio-temporal maps of the rivers using Gaussian processes, Kalman Filters , or other techniques.

On July 3rd, 2018, we made an expedition to Bakersfield, CA to survey a section of the Kern River. The ASV was deployed with RC controls and surveyed a section of river while collecting velocity and bathymetry data. A bathymetry map of the surveyed section was then created using a 1D Kalman filter. We are currently working on setting up autonomous navigation and path planning for the ASV.