Virtual RobotX 2019
The power of ROS and Gazebo to drive the development of robotic systems and improve robotics education was on full display in Singapore late last year. Open Robotics worked in conjunction with the Naval Postgraduate School and the Office of Naval Research presented the 2019 Virtual RobotX Competition results at the RobotX Interactive (RXI) Forum. The RXI Forum brings together academics, students, governments, and industry professionals, to discuss the application of robotics to maritime environments and this year the event included poster sessions, hackathons, demonstrations, discussions with government officials and industry veterans, and most importantly an announcement of the Virtual RobotX competition results. This international event draws attendees from around the globe, and this year saw a diverse set of student competitors from universities in countries such as the United States, Canada, Singapore, South Korea, Taiwan, and Australia.
The Virtual RobotX Competition uses ROS and Gazebo as a competition platform where students are given a virtual Unmanned Surface Vehicle (USV) and a set of maritime tasks to perform. To make the competition realistic, Gazebo is configured to simulate a number of maritime phenomenon like strong winds and a wavy ocean environment, everything from “mild to wild” is possible. Each Virtual RobotX competitor was given a virtual USV and a sensor payload and then tasked with building the algorithms and behaviors necessary to perform a variety of tasks. The virtual USVs closely resemble the WAM-V ASV which is used by a number of groups for USV research. Given the complex maritime environment students could configure up to four thrusters on their USV in any way they saw fit, however there were four nominal configurations to get students started. For the 2019 competition the nominal sensor payload included a differential GPS device, a suite of three high resolution cameras, a high-fidelity inertial measurement system, and a number of of potential LiDAR sensors. Students were allowed to modify and extend the sensor payload to suit their needs.
Once the teams had configured their USV they then had to write software to autonomously perform half a dozen tasks. For Virtual RobotX, the tasks represented high-level components that could be composed into more sophisticated autonomous missions. These tasks, while they appear simple, can become incredibly difficult in marine environments due to the effects of the wind and waves which can push the vehicle around at any point during the task. In order to evaluate the performance each task had an associated scoring metric which was used to calculate the competition winner. The tasks included:
Station-Keeping - hold a given position on the water despite being moved by wind and waves. This task was scored by how far off the USV was from the desired position.
Wayfinding - navigate the USV through a series of predefined GPS waypoints. This task was scored by the difference between the desired route and the actual route.
Perception - for this task the USV was shown buoys with different colors/shapes and scored based on correct identification and localization accuracy.
Navigation Channel - navigate a path, demarcated by colored buoys, while avoiding obstacles. For this task the score was determined by the total time required and correctness.
Dock - for this task the USV must maneuver itself into a dock. Points were awarded based on if the task was completed.
Scan-and-Dock - dock the USV by first locating the correct slip using a colored flag, and then dock. This task was scored similar to docking, but with added points for identifying the correct slip.
The video below shows and example of each of the tasks and shows a nominal USV configuration.
[vimeo 384618407 w=426 h=240]
This year’s competition had teams from eleven universities from across the globe. For the final competition each team repeated every task six times with varying operational parameters and initial conditions. The final scores for each team were calculated using a fairly complex scoring mechanism but the scores roughly correlate to the average score for all of the six runs. The competition was close, with no one team dominating on every challenge. The final scoring charts can be found here. The results of the competition are as follows:
1st Place: Korea Advanced Institute of Science and Technology (KAIST)
2nd Place: Georgia Institute of Technology
3rd Place: The University of Sydney
4th Place: National University of Singapore
5th Place: University of Florida
6th Place:
Embry-Riddle Aeronautical University
Flinders University,
Florida Atlantic University
McGill University
National Taiwan University
University of Hawaii at Manoa
Open Robotics would like to thank the Naval Postgraduate School and the Office of Naval Research for making RobotX possible. With our experience with the DARPA Robotics Challenge, the NASA Space Robotics Challenge, the Toyota Prius Challenge, and the current DARPA Subterranean Challenge we know that these types of simulated and open source competitions create a fun, relevant, and accessible platform for students and researchers to accelerate fundamental engineering and science research and we look forward to future events.