Cooperative data collection mission

Cooperative data collection mission

Coordinated collective acquisition:

• Maximum coverage of the study area at different scales.
• Cooperate to allocate acquisition zones and share resources: robots, sensors, etc.

Advantages:

• Increase the quantity and variety of data collected.
• Encourage collaboration and exchange between participants.
• Develop synergies and multidisciplinary approaches.
• Enable more detailed and precise analyses.

How can you get involved?

• Contact the organizers for more information submeeting@univ-tln.fr.
• Prepare your equipment and data collection software.
• Spend a few minutes at the end of the dive collecting data at your chosen site.
• Share your data and results with other participants.

In short, cooperative image data collection is a unique opportunity to pool resources and skills to create a rich database useful to all.

Don’t hesitate to join the adventure!

Data acquisition and transmission guide

After each dive, you’ll have access to external hard disks to store the data you’ve acquired. Each disk will have the same organization:

• Site of interest 1
  • Robot A
  • Robot B
  • …
• Site of interest 2
  • Robot A
  • Robot B
  • …
• …

First, go to the folder [site of interest you just visited]/[the robot you used].

Next, create a folder whose name is the approximate date of the visit in the format YYYYMMDDHMMSS, with YYYY the year (2022), mm the month indented (04 for April), dd the day indented (07 for the 7th), HH the hour indented (14 for 14h/2PM), MM the minutes indented and SS the seconds indented. For example, if your visit took place on April 7 at approximately 2:19 p.m., the folder name will be 20220407T141900.

In this folder, you can store all your data, whether rosbags or videos. If you only have video files, please name the videos according to their recording time in the same format as the folder. For example, if your video was recorded at 9:05 am on April 8, the video file should be named 20220408T090500.mp4 (depending on the video extension).

Hard disks will also contain a folder named Calibration with the following organization:

• Calibration
  • Robot A
  • Robot B
  • …

In the folder corresponding to your robot, you will be asked to deposit the data acquired during the calibration session.

3D reconstruction challenge

After data collection, data processing!

The database will be populated as the workshop progresses, and as soon as the first data is available in the database, it will be possible to start processing. Everyone will be able to participate, even those who don’t collect data! The aim is to use the image data to produce local and global 3D reconstructions of the area covered. Any additional value-added processing will be of interest (location of shots, analysis of day-to-day data completeness to guide subsequent acquisitions, pre-processing, mosaics, object detection, etc.). Here again, guidelines will be provided on how to share these results with other participants.

Based on certain acquisition sequences, we will strive to establish a ground truth of 3D reconstruction that will enable participants to quantitatively compare their algorithms. Since it would be too complicated to compare results obtained with different robots equipped with different sensors and not synchronized with each other, the 3D reconstruction challenge is likely to be restricted to acquisitions made by a single robot. However, the processing of acquisitions made with other robots will enable a qualitative analysis of the results obtained.

The reconstruction challenge will continue for a few months after the end of the workshop.

At the end of the challenge, we hope to write a collaborative scientific article to present the database and initial results with the people involved in data collection and processing.