Tuesday, April 28, 2020

Volta Cohort Pitch Competition

Back in March, our controller at Measurand mentioned a competition that I might be interested in: The Volta Cohort Pitch Event (https://voltaeffect.com/cohort-finalists-spring2020/). There was a written online application with a number of questions pertaining to your business. 15 applicants would be picked to pitch their business at an event in Halifax on May 13, and up to 5 of those would win investments of $25000 and become part of the Volta cohort where they would receive mentoring and assistance with their business. Last week an email arrived to say that In Nature Robotics had made it to the group of 15, although due to  COVID-19 the pitch event will be held entirely online. So Jata and I will have to figure out what to tell people about the company in a 3 minute online PowerPoint spiel. No problem!

Last week was also favorable for testing, although not at first. On Saturday I drove down to the same launching point as the week before, but it was much windier, and the current looked even stronger. While I was waiting for a GPS signal and making some adjustments to the protective cage, the GPS signal actually came through and the propeller whirred to life, knicking two of my knuckles. I had considered moving to a beefier more powerful propeller, but now I'm glad I didn't! Here is a short video recording of the current, which doesn't really do it justice:

I had wanted to try out some new software that had just been coded that morning. As sometimes happens, there was a tiny bug that ruined the entire test. I had forgotten to convert a heading variable from radians to degrees, which totally messed up the new algorithm to skip waypoints that the boat had already passed. So after the boat had struggled against the current for about an hour I gave up and towed it back. 

Sunday was much better. It actually would have been nice to have some strong wind and current to really test out the fixed software, but the current seemed a bit less, and there was scarcely any wind at all. It took the boat about 2 hours to go from the launching point down to a point a little beyond the Westmoreland Street Bridge, a trip of about 9 km. On the way it collected pH and temperature data, and took a number of pictures:

The route AMOS followed (I towed it upriver along the shore).

Temperature Data (with interpolation turned on)

pH Data (probably not accurate, I'm going to try cleaning the probe before the next test).

One of the planned pictures was this one of the Delta Fredericton.

Zoomed in shot of some guy in a kayak.

Nice shot of Westmoreland Street Bridge.

It took about 2 hours to tow AMOS back to the launching point, and since it was a nice day many people were out walking and biking. Quite often they would ask what I was towing and one lady even took my picture. I did my best to give them a short explanation; it was probably good practice for next month's competition. 😊

Tuesday, April 21, 2020

How to Run Down The River

This past weekend we were blessed with some nice warm, sunny weather so it seemed like an ideal time to try out the new AMOS-Cat boat on something more significant than the large puddle in our backyard. On Saturday, I biked down to the St. John River to check out a prospective launching area about 8 km upriver from the main downtown core of Fredericton. The spot had a bit of a parking area for 5 or 6 cars and a nice inclined path leading down to the water. The current looked a bit fast, but not so fast that I couldn't tow AMOS upriver in the kayak if necessary.

So I mapped out a course that would take AMOS about 8 km down to the Westmoreland Street Bridge, collect some sensor data and collect pictures at selected locations en route:

On Sunday AMOS started the course shortly after noon. There was the usual 10 or 15 minute wait for a valid GPS signal (a new GPS is on its way which should hopefully speed that up) and then it was off down the river. The strong spring current pulled it along rapidly at first, but it tended to overshoot the interpolated waypoints (there are 10 interpolated waypoints between each of the yellow markers shown above) and then reverse direction to try to get back to the waypoint that it had missed. The software has an algorithm which gives up on trying to reach a particular waypoint (and moves on to the next one) if the distance to that waypoint increases over any 4 minute interval. The problem on this day though was that in most places AMOS couldn't make any forward progress against the current, so it spent a lot of time just "hovering" out in the middle of the river, or even moving backwards in areas of strong current. I will need to change the algorithm to move onto the next waypoint whenever the direction to that waypoint is opposite to the intended direction of the planned route. That will result in some loss of waypoint achievement accuracy, but it should be preferable to waiting for long periods of time while the boat is barely moving against a strong current.

AMOS made it about 2 km down the river before I gave up and decided to tow it back upriver with the kayak. This was about 45 minutes of hard paddling! Here is a map of the route that AMOS took:

The return trip was much closer to shore to avoid as much of the current as possible and because the water was still freezing cold.

Lots of people were out walking on the nearby river trail, and many of them asked about the robot (safely from a distance) so I guess it was good publicity, even though it wasn't quite functioning as intended. Here is a short video taken of AMOS moving upriver in an area where the current wasn't too bad:

Tuesday, April 14, 2020

Backyard Puddle Test

We had a fair amount of rain yesterday and last night, which combined with the leftover snow from winter to make a giant puddle in the corner of our backyard. Although it didn't really provide quite enough room for AMOS to get going, I felt the need to try out the boat in  some actual water for the first time this year:

The boat quickly became fetched up on an alder and needed to be rescued by wading part-way into the puddle and reaching for it with a hockey stick. This version of AMOS (AMOS-Catamaran or AMOS-Cat for short) seemed to have about the same draught as the surfboard version, i.e. about one cm, maybe a bit more at the back end due to the weight of the battery.

A lot of work was done this past week on the new version of the PC BoatCaptain software for controlling AMOS and viewing info on ArcGIS maps. I was able to get the boat route, safety locations, and photographic locations all to appear properly in the application and in the saved web maps, and was also able to load those saved web maps into the application. For some reason though, I could only view multi-colored data points from within the application; when I tried to save a map with multi-colored data points in it to the web, all of the points in the web map became the same color. I suspect it's just a limitation of the SDK that I'm using, or the web service that does the saving of the map, but I asked a question on a relevant forum to see if anyone else knew.

For this week I want to focus more on the new BoatCaptain software for controlling and interacting with AMOS. The weather is quickly warming up and it will soon be time to start collecting more test data!

Tuesday, April 7, 2020

Happy Birthday to That Hobo Who Works in Our Backyard Playhouse

There is a bit of a running family joke that there is a hobo squatting in our daughter's playhouse. I have been using the playhouse as a work space lately in the afternoons when the sun is out. It is nice and quiet there, without any of the distractions that are typical within a 5-person home. Today was my birthday, so unbeknownst to me, Kelly went out in the morning to decorate it before my afternoon work session:

Much of this past week has been spent re-writing the graphics code for the new version of the desktop BoatCaptain software. I eventually figured out the methodology required for creating graphics in the application's map that will also show up the same way in a version of the map stored on the web. Saving simple symbols was fairly straightforward, but getting custom image icons (using the "PictureMarkerSymbol" .Net ArcGIS class) to save properly in the version stored on the web took me hours to figure out. It turns out that you need to explicitly define the dimensions of the PictureMarkerSymbol, otherwise it won't display at all on the web.

For example:
      PictureMarkerSymbol crossMarker = new PictureMarkerSymbol(new Uri(sImagePath));
      crossMarker.Height = 20;
      crossMarker.Width = 20;
Kelly happened to notice a Facebook post about a boy that had 3D-printed a number of "ear guards" that help to ease some of the pressure on the ears of hospital workers wearing masks with elastic straps that hook behind the ears. The model of the part can be found here: https://www.thingiverse.com/thing:4249113. She wondered if I could make any on my printer, and I wasn't sure; I usually have trouble printing things with small details over a large surface area. After a bunch of tweaks and a few failed prints I was able to find some settings I had never tried before that worked really well. I made 3, which she gave to a nurse she knows, so we'll see if they're actually useful or not. The new 3D printing configuration also worked really well for this pH probe holder that I just made today: