DATASETS | AUV Lab

MARINE PERCEPTION

The MIT Sea Grant Marine Perception Dataset is a public dataset for autonomous surface vessels developed by the team in the Autonomous Underwater Vehicles Laboratory (AUV Lab). We collected a large multi-modal sensor data set for mobile robotics research in the marine domain. The data set was collected by a mobile marine autonomy laboratory, built upon a modified Boston Whaler marine vessel. The dataset consists of multiple scenes to show the diverse complexity that one might expect when operating a marine vessel in a coastal region. Data was collected on the Charles River, Boston, MA out into Boston Harbor to capture the data provided by a recreational river full of sailboats and kayaks to the many ferries, barges, and water taxis in a highly trafficked harbor.

SENSORS USED

The dataset contains time stamped data from the following sensors:

VLP-16 Velodyne 3D Lidar
4G Broadband Radar
3 FLIR Blackfly 1.3 MP Cameras
2 FLIR ADK 0.3 MP Infrared Cameras

2020_Philos-sensor-suite-sensor-coordinates

DIVERSITY OF CONDITIONS

Sailboat Traffic

Harbor Traffic with Snow

Bridge

Harbor Traffic with Rain

MISSION MAP

This map illustrates typical missions around the Charles River Lower Sailing Basin (red), in the vicinity of Poorman’s Landing (blue), and transits along the Charles River to/from Boston Harbor (green). All points are from GPS log data, so some GPS points are missing when GPS becomes unavailable.