Kenichi Fujita, Yuya Hamamatsu and Hiroya Yatagai (The University of Tokyo)
University of Tokyo students challenged the SAUVC for the first time supported by OES Japan student program. Enjoy the story!
Singapore Autonomous Underwater Vehicle Challenge 2019 was held at Singapore Polytechnic on 8-11th March 2019. The event has been held annually since 2013, and this is the first time for a Japanese team to join this challenge.
This report provides an overview of the event including what we felt during the competition, especially about the difference between SAUVC and an underwater robot competition held in Japan, which we had attended.
2. Introduction of SAUVC
SAUVC has three rounds: video audition, qualification round, and final round.
In the video audition, teams have to upload a short video to show their AUV actually works. This year, this audition was conducted around December 2018. 61 teams registered, and 40 teams, including our team, passed this round.
The next round is the qualification round. In this round, every team tries to make their AUV pass the gate 10m apart. The teams are ranked by the time to pass the gate.
Top 15 teams qualify for the final stage.
The final round has the following 5 tasks (Fig. 1):
(1) Navigation (10pt)
(2) Target acquisition (50pt)
(3) Target reacquisition (60pt)
(4) Localisation (40pt)
(5) Other bonuses (Time, size, weight) (52pt)
The AUV passes through the 1.5 m height gate (the color of the left pole is red and the other pole is green). The AUV has to complete this task before proceeding to other tasks.
(2) Target Acquisition
Four drums are installed at the bottom of the pool. One is blue and the others are red (one red drum has a pinger inside). Dropping a ball in the blue drum gives 30pt, in the red drum with pinger gives 50pt, and in any other red drum gives 10pt.
(3) Target ReacquisitionAfter the task of Target Acquisition is completed, the AUV can try to pick up the ball in the drum. If the AUV reacquires the ball, the team gets the score.
The AUV localizes a pole called “flare,” marked with a pinger, located somewhere in the main arena and bumps it to drop the golf ball on top of the flare.
3. Our Team
We had joined the AUV competition held in Kobe, Japan, in May 2018 (OCEANS’18 / Techno-Ocean 2018 Kobe (OTO’18)). The tasks are swimming through a gate, touching a buoy, coming close to a pinger, dropping a ball, and landing on a plate.
Our vehicle ‘Minty-Roll’ successfully completed 3 tasks, swimming through the gate, touching a buoy, and dropping a ball (Fig. 2).
Three teams from The University of Tokyo, Kyushu polytechnic college, and Kyushu Institute of Technology joined, and our team got the first prize.
In October 2018, we heard OES Japan Chapter plans to give a scholarship for student teams to join SAUVC. Therefore, we applied and were successfully accepted.
To join SAUVC, we had to modify both hardware and software of the vehicle. In hardware, we added a kill-switch and changed the batteries. The size of the battery was too big to carry in airplanes, therefore we modified the AUV to use smaller batteries. In software, as we decided to focus on passing the gate and detecting the flare (because completing all the tasks seemed to be difficult for us), codes for the tasks were rewritten based on the previous program used for detecting a buoy at the competition in OTO’ 18 (Fig. 3).
4. Our Struggle in Singapore
(1) Preparation Day (Fig. 4)
The first day of SAUVC had short guidance and preparation for qualification and final round. After giving a short presentation on our AUV to the judges, we moved to the pool and checked our AUV. We were worried that some parts could be broken during transportation, although it looked like no problem. Through the number of trials, we found the problem that one of four thrusters did not work well and the AUV could not go straight, once in ten times. We tried to detect the cause, but we could not.
(2) Qualification Round (Figs. 5, 6)
Our time slot was in the early morning. In our first and second trial, one thruster did not work well and our AUV swam to a different direction. The third trial was the most regrettable. Thrusters had no problem and our AUV swam straight to the gate. However, our AUV was slower than we expected. The AUV was programmed to move forward for 100 seconds. The AUV could not reach the gate before the AUV’s thrusters stopped. If we programmed to move for a longer time, the AUV would have passed the gate. The fourth trial was the same as the first and the second. Finally, the AUV could not pass through the gate, and we could not go to the final stage. Even if our AUV passed the gate, we would not be one of the top 15 teams, considering the time.
5. Differences Between the Two Events
We joined competitions both in Japan and in Singapore. We noticed there are some differences between the two competitions. In this section, the differences are discussed in the context of AUV, team, rule, and atmosphere.
One of the biggest differences is the AUV’s size and speed. The AUVs in SAUVC were smaller and faster compared with those found in OTO’ 18. The size of our AUV is 0.67 m (height) × 0.40 m (width) × 0.46 m (length). This was the smallest in OTO’ 18, but this is one of the biggest in SAUVC. The weight is also the same. Ours is 15 kg, but I heard the weight of many other AUVs in SAUVC is around 5 kg. One of the reasons seems to be that the AUVs have to move as fast as possible in SAUVC to pass the qualification round. More than half of the vehicles looked to use thrusters from Blue Robotics (T100 or T200) . Because the thrusters are the same, smaller and lighter body is beneficial for gaining speed.
Another difference is the variety of the shape (Fig. 7 Left). The box-shaped AUV, from Far Eastern Federal University, is the champion of SAUVC 2019. The distinctive design attracted the attention of many participants. It is designed to be able to efficiently accomplish the tasks of the final round using sensors, such as a hydrophone and a camera, without sacrificing mobility. The center of gravity is placed to be lower for higher stability in water. This vehicle passed the qualification round at a relatively higher rank, and it also accomplished some tasks in the final round.
Another interesting vehicle is a UFO-like cylindrical AUV (Fig.7 Right) from National University of Singapore (NUS). The most distinctive structure is the heave thruster placed at the center of the cylinder. They also used a stereo camera for object detection and navigation. Though we heard the system did not work well, if it works it would bring great advantages for accomplishing tasks.
Our AUV was made aiming for robustness enough for the shallow sea. This is beneficial for winning Japanese competition and for research.
There was a big difference in the number of teams and team members. In OTO’ 18, there were only 3 teams (10, including freestyle category), but SAUVC had around 40 teams. Some participants in SAUVC were surprised to hear Japanese AUV competition only had 3 teams. Some teams in SAUVC had more than 10 members and averaged 5 members. Our team has only 3 members, which is almost average in OTO’ 18. We were asked many times about how to build and operate the AUV with such a small team. Our answer was “we do everything.”
We found almost all the other teams in SAUVC are club teams. For example, Bumblebee, the team from NUS, is a robotics club participating in international competitions such as Robosub  and Robot X  besides SAUVC. In OTO18, two of the three teams consisted of graduate students. Many Japanese universities have club teams for robotics competitions, but their main focus is flying a robot like a drone or robots working on the ground.
The recommended way of localization is different. In OTO’18, the recommended method for localization is line tracing. On the other hand, the main arena in SAUVC has no line to follow. Therefore, the main localization method was pinger and object detection.
Time to pass the task, the weight of the AUV are not important in Japanese competition. The weight is measured just to check that the vehicle is not heavier than a certain threshold. Therefore, we did not consider the speed to be important. We thought that we can go to the final round if the AUV passed the gate. This was not true. We were required to pass the gate much faster in the qualification round. Although we planned to use a vision-based gate detection algorithm, some of the teams that passed the qualification round seemed not to use any visual feedback. Their AUVs seemed just to go straight using PID control based on gyro or compass.
Underwater scenes were displayed in real time by underwater cameras in SAUVC (Fig. 8). This was very effective, as it is often difficult to understand what is happening underwater because the underwater environment is not clearly seen from outside. In Japan, the chairperson tried to verbally explain the situation, but sometimes it is difficult.
We had a lot of chances to communicate with other teams in both events. We were able to move around the venue and ask questions about other AUVs. All the teams gathered at the end of the competition. That also helped our communication.
Almost all the teams in SAUVC had some SNS accounts. As the accounts have many followers, they are a good platform for advertisement. For some teams having sponsor companies, advertisement is especially important. SAUVC offered SNS award for the team which earned the highest number of “likes” during the event. Surprisingly, the winner earned 46,805 likes in 3 days. Japanese AUV community is not so big. SNS seems to be a good way to attract young people who are interested in underwater robots.
We learned a lot from SAUVC 2019, although we could not pass the qualification round. This is the first time for a Japanese team to join the event. There were some differences between Japanese competitions and SAUVC. We hope that more teams from Japan will join future SAUVC events. Also, we hope more people, especially more students, join the world of underwater robotics.
This project was supported by Youngster Robocon Support Program 2018 offered by IEEE/ OES Japan chapter. We would like to express our sincere appreciation to the organization for strong support.
 H. Horimoto, T. Nishimura, T. Matsuda, AUV “Minty Roll” and results of “Underwater Robot Convention 2017 in JAMSTEC”, IEEE OES Beacon Newsletter, 6(4), 77–79 (2017.12)
 Blue Robotics—ROV and Marine Robotics Systems and Components, https://www.bluerobotics.com/ (accessed 2019-05-07)
 RoboSub | Robonation, https://www.robonation.org/competition/robosub (accessed 2019-05-07)
 Maritime Robotx Challenge | AUVSI Foundation | Hawaii Robotics Competition 2016, https://www.robotx.org/ (accessed 2019-05-07)
 Singapore AUV Challenge, https://sauvc.org/ (accessed 2019-05-07)
 Underwater Robots Competition, http://www.oceans18mtsieeekobe.org/underwater-robots-competition/ (accessed 2019-05-07)