Results

2nd RAMI Cascade Campaign

for Marine Robots

The Second RAMI Cascade Evaluation Campaign for marine robots has ended, it challenged the navigation and perception capabilities of participant teams by presenting them training datasets collected on scenarios similar those of physical campaigns.

Based on these data, teams were required to develop their software and tune their algorithms, which were then evaluated on a test dataset composed as follows:

- 4  Images showing colored buoys (2 Red, 2 White, 1 Yellow)

- 3 Images showing Black numbers on yellow pipes

- 4 Images showing Black numbers over a red background

- 2 Images showing a Red marker over a yellow pipe

- 3 Images showing manipulation console objects

- 4 Images showing only tubes and water with no OPIs

The following 4 teams registered to the competition:

Team name Institution/Company Country
Aqua Squad Virginia Tech US
Kadal-AI Indian Institute of Science (IISc), Bangalore IN
OUBOT Obuda University HU
Team ERGO - UNIPI University of Pisa IT

Every team submitted a Docker Image as a Docker Hub repository, which was evaluated following the rules described here

Ranking

The competition was won by Team ERGO-UNIPI with the following results:

Total:

  • 1st: Team ERGO-UNIPI with a score of 123/149 (82.6%).

The ranking for the different functionalities is:

Classification:

  • 1st: Team ERGO-UNIPI with a score of 51/55 (92.7%).

Identification:

  • 1st: Team ERGO-UNIPI with a score of 27/43 (65.9%).

Localisation:

  • 1st: Team ERGO-UNIPI with a score of 45/51 (88.2%).

Organising committee

Chair

Dr Gabriele Ferri (CMRE, Italy)

Co-chairs

Alessandro Faggiani (CMRE, Italy)

Dr Tommaso Fabbri (CMRE, Italy)