The next version of the Aqua underwater will be getting a workout in the McGill pool sometime soon with new legs. We have some problems with the calibration of the intertial measurement unit, but otherwise things are proceeding apace. The amphibious legs, which allow both walking and swimming to be accomplished efficiently are looking really good and this should be an exciting test.


(The legs shown here are the older swimming flippers and not the new still-confidential version 3 legs. Movie here.

These legs combine compliance and swimming factors to greatly exceed the performance of our prior legs. They have been on the drawing board and in prototype form for some time, and it will be important to really assess their performance accurately.

After several years of successful underwater robotics experiments we had our first case of water leakage into the inside of the robot. As one might expect, it was due to human error -- a miscommunication regarding who was responsible for tightening certain seals. As a result, water from the McGIll pool leaked into the robot vehicle during a test and quite a lot got in before it was discovered. THings seems more-or-less OK; this is the kind of problem you want to have in frash water, not salt water. As such, it may have an up-side getting rid of complacency and reminding everybody to be fanatically careful when the robot is at the real sea trial.

Thinsg are still being dried and re-assembled, but it could have been a lot worse.

FOLLOWUP: All this stuff was repaired and worked in later pool trials.

We had the first tests of Ramius, the premiere model of version 2.0 of the aqua robot platform, in the open ocean today. The weather in Barbados has been quite rainy, which is unusual for early January, but not exceptional. As a result, we had two major problems: the water visibilty was moderately poor, and the people on land monitoring the tests by computer had lots of trouble with the risk of rain hitting the laptops. As a result, we had a fairly bad weekend, even though we did get some useful data.

There are also some broken wires that developed inside the robot, but those have been fixed all all systems are "go", although we are not almost a day behind schedule. In addition, the silt from the rain can takes days to dissipate, and there is more rain in the forcast.

Despite the problems, there are some exciting tests scheduled for tomorrow including a deep water validation of the Kroy model of Aqua 2.0. If all goes well, we'll take it down to 120 feet at the Stavronikita wreck at the South end of the island. Even in the rain we can run Ramius off the shore with the land users inside a covered boat house. (These tests are the subject of a future blog entry.

While doing the tests,we also completed the rescue diver course we started back home. The skills should be useful for finding lost robots, as well as divers in distress.

News on the amphibious robot we have dubbed Ramius: version 2 of the AQUA platform. I had previously noted that we were having some problems with the hardware and the weather. The weather eventually improved and while the visibility has remained corrupted, the tests went ahead successfully. We used 3 scuba dive teams to ferry the robot down to 120 feet, take stops along the way to calibrate the strain guage used for depth measurement, and had one team devoted to keeping an eye on the other cognitively loaded diving team(s). The procedure was carefully planned and included precautions in case the robot imploded, or for ditching the gera in case of emergency. After all that, the whole thing came and went very quickly.

We also validated some important new hovering and station keeping behaviors by the robot: various ways of staying on one spot and keeping an eye on things. Some behaviors didn't work well, or couldn't be properly tested. The sub-optimal weather not only hindered visibility, but also led to a high rate of sea sickness. The rate at which we lost fiber optic cables was a bit alarming, but the core requirements of the test went well.

Nicholas was part of the St. George's School team participating in the CRC Robotics competition called Archemedia 2007. They teams built robots that have to haul 1 foot diameter rings around an arena and stack them up. Their team came in second place overall, as well as doing well in several sub-categories. This includes having a kiosk (booth) that placed in the top 6, for which I believe Nicholas was in charge. This kiosk featured a huge metal dragon (the school emblem) on a pneumatic cylinder that allowed it to go from floor level to about 30 feet in the air. The dragon blew smoke too. The background of the kiosk featured pictures of the team members and biographical information. In addition to a kiosk and a robot, the teams also prepared a journal documenting their activities, a video and a web site.


(click to enlarge)


One of the striking things about the competition is that is has the hype of a college football game. This includes large stands with screaming spectators and plenty of jumping around. Many other schools were involved including a technical school for adults, an assortment of private schools, CEGEPs (junior colleges) and some huge public high schools. Other notable features were a kiosk built by ECS School in which the students dress as mermaids, including having costumes with a single lag, an impressive robot from Laval that could pick up many rings at once (see below), and a kiosm constructed from water bottles. In this latter case, the audience was challenged to guess how many bottles the kiosk was constructed from. [ I guessed 3782, was within 14 of the right answer, and it won me a huge pile of candy. ]

Overall this kind of event is great promotion for science and engineering.

The Robotics Science and Systems (RSS) conference is on track for Atlanta, GA this summer. It will be adjacent to RoboCup so should be able to attract a range of participants. There had been a chance of attracting RSS to McGill and Montreal, but that would have represented a huge amount of work (for me!) and Georgia Teach seems to have superb infrastructure and financial support for this kind of thing. The big news is that RSS seems to have had a substantial increase in submissions this year, so the conference seems healthy and it promises to be exciting. Reviewing is in progress, and official figures should be available soon.

RSS is a key conference for hard core scientific results in all of robotics, from mapping algorithms to mechanical design. It features a single track of 30 to 60 minutes presentations of original scientific results. RoboCup involves a series of robot competitions from version seriosu teams. These include physical robots an software simulations. The teams that compete are typically composed of graduate students and/or professional who are pushing the boundaries of what can be achieved.