DATASETS | AUV Lab

Philos – Target & Sailboats

This dataset was collected on 2019/10/29 on the vessel R/V Philos on the Charles River near the MIT Sailing Pavilion, Cambridge, MA. Skies were overcast and the weather was mild. The overall intent of testing that day was detection via Lidar and avoidance of a large (~1 m by 2 m) anchored block of Styrofoam while capturing other sensor data. During testing, several sailors passed into and out of the test area, including one that passed close to the target. Duration of the dataset is approximately 00:03:25 (epoch 1572379229 to 1572379426.)

HIGHLIGHTS

Sailors are readily visible in infrared video
Radar and Lidar pointcloud data can be displayed in the same coordinate frame via Rviz, for example, with radar Decay time = 2.5 s and Size = .1 and Lidar Size = .09. Note that sailboat lidar outlines appear to travel ahead of the related radar pointclouds, which refresh every 2.5 s.
Good left and right video acquired at 12 fps and infrared at 30 fps
Broadband radar channel 0 (initially 50 m, changing quickly to 500 m) and channel 1 (range 100 m) capture nearby sailboats, target Styrofoam, and Sailing Pavilion dock complex.

Dataset ID: philos_2019_10_29__16_00_19_target_and_sailboats

Filename: philos_2019_10_29__16_00_19_target_and_sailboats.tar.gz (6.58 GB)

NOTE ~ Downloads are not possible via Chromium-based browsers at this time.

Philos – Mooring Field

This dataset was collected on 2019/10/30 on the vessel R/V Philos in Boston Harbor near the New England Aquarium, Boston, MA. Skies were overcast and the weather was mild. The overall intent of testing that day was collection of typical harbor features. During testing, imagery of sailboats in a mooring field was captured, as well as a ferry preparing to dock. Duration of the dataset is approximately 00:02:27 (epoch 1572454214 to 1572454362.)

HIGHLIGHTS

Radar and Lidar pointcloud data can be displayed in the same coordinate frame via Rviz, for example, with radar Decay time = 2.5 s and Size = .1 and Lidar Size = .09. Note that sailboat lidar outlines appear to travel ahead of the related radar pointclouds, which refresh every 2.5 s.
Good left and right video acquired at 12 fps and infrared at 30 fps
Broadband radar channel 0 (range 500 m) and channel 1 (range 100 m) capture nearby sailboats and ferry.

Dataset ID: philos_2019_10_30__12_50_09_mooring_field

Filename: philos_2019_10_30__12_50_09_mooring_field.zip (2.92 GB)

NOTE ~ Downloads are not possible via Chromium-based browsers at this time.

Philos – Lighter 1

This dataset was collected on 2019/12/11 on the vessel R/V Philos in Boston Harbor between Georges Island and Hull, MA. Temperature was near freezing. The skies were just clearing after a light snow squall. The overall intent of testing that day was collection of typical harbor features. During testing, a fuel lighter and tug were captured while Philos was on plane. Duration of the dataset is approximately 00:01:11 (epoch 1576081550 to 1576081620.)

HIGHLIGHTS

Good left and right video were acquired at 12 fps and infrared at 30 fps
Left and right video show close approach to the lighter and tug at speed and a
banking turn to port
The infrared video clearly displays structures heated by tug and lighter exhaust
Snow on the left and right camera lenses partially obscure the frame
Lidar does not capture anything notable in this dataset
Although broadband radar channel 0 (range 100 m) and channel 1 (range 500) capture the lighter and nearby landmasses at 24 fpm, radar imagery is noisy. This highlights the importance of adjusting radar to minimize noise before and during each run and the difficulty of doing so in adverse weather with minimal staff.

Dataset ID: philos_2019_12_11__09_58_25_lighter1

Filename: philos_2019_12_11__09_58_25_lighter1.zip (1.19 GB)

NOTE ~ Downloads are not possible via Chromium-based browsers at this time.

philos_2019_12_11__09_58_25_lighter1-camera-with-snow

Philos – Lighter2

This dataset was collected on 2019/12/11 on the vessel R/V Philos in Boston Harbor in Hull Gut, between Hull, MA and Peddocks Island. Temperature was near freezing. Skies were clearing just after a snow squall. The overall intent of testing that day was collection of typical harbor features. During testing, a fuel lighter and tug were captured while Philos was on plane. Duration of the dataset is approximately 00:02:20 (epoch 1576082065 to 1576082204.)

HIGHLIGHTS

Good left and right video were acquired at 12 fps and infrared at 30 fps
In the last 10 seconds of video, Philos makes a hard 180o turn, sweeping past the lighter and tug
Snow on the left camera lens partially obscures the video frame
Lidar does not capture anything notable in this dataset
Although broadband radar channel 0 (range 100 m) and channel 1 (range 500) capture the lighter and nearby landmasses at 24 fpm, radar imagery is noisy. This highlights the importance of adjusting radar to minimize noise before and during each run and the difficulty of doing so in adverse weather with minimal staff.

Dataset ID: philos_2019_12_11__09_58_25_lighter2

Filename: philos_2019_12_11__09_58_25_lighter2.zip (2.61 GB)

NOTE ~ Downloads are not possible via Chromium-based browsers at this time.

Philos – Long Island Bridge

This dataset was collected on 2019/12/11 on the vessel R/V Philos in Boston Harbor near the site of the former Long Island Bridge, Quincy, MA. Skies that day started snowy, then transitioned to snow haze, finally clearing. Air temperature was near freezing. The intent of testing that day was capturing data in snowy conditions. Duration of the dataset is approximately 00:03:30 (epoch 1576080900 to 1576081109).

HIGHLIGHTS

A high-speed ferry passes through the bridge abutments just as Philos approaches. Philos takes the ferry wake as it passes the abutments.
Good left and right video at 12 fps and infrared (ferry visible) at 30 fps
The ferry and abutments are visible in the lidar.pcap file
Although broadband radar channel 0 (range 100 m) and channel 1 (range 500) capture the bridge abutments and ferry at 24 fpm, radar imagery is rather noisy. This highlights the importance of adjusting radar to minimize noise before and during each run, and the difficulty of doing so in adverse weather with minimal staff.

Dataset ID: philos_2019_12_11__09_58_25_long_island_bridge

Filename: philos_2019_12_11__09_58_25_long_island_bridge.zip (3.35 GB)

NOTE ~ Downloads are not possible via Chromium-based browsers at this time.

REx 4 – Harvard Bridge

This dataset was collected on 2019/10/08 on the vessel R/V REx 4 at the Harvard Bridge over the Charles River near the MIT Sailing Center, Cambridge, MA. Skies were overcast. The overall intent of testing that day was collection of sensor data during autonomous navigation of the bridge, including operation in GPS denied areas. During testing, REx passed under the bridge twice upriver and twice downriver. Duration of the dataset is approximately 00:04:20 (epoch 1570556012 to 1570556272.)

HIGHLIGHTS

Good center video was acquired at 12 fps and infrared at 30 fps
Video shows multiple passes under the bridge
Lidar shows details of the bridge structure, including partial details of the superstructure. Note: The lidar in the PCAP file is rotated about the Z axis -90°. Original and corrected topics are available in the rosbag file.
Broadband radar channel 0 (range 250m) and channel 1 (range 100m) captured transits under the bridge, including limited portions of the bridge abutments and superstructure. Since radar images are XY only, these bridge elements combine to appear as solid obstacles.

Dataset ID: rex_2019_10_08__13_29_51_harvard_bridge

Filename: rex_2019_10_08__13_29_51_harvard_bridge.zip (4.87 GB)

NOTE ~ Downloads are not possible via Chromium-based browsers at this time.