Monday, August 29, 2011

A lake on an island on a lake on an island.

Today's post is centered on lakes and islands that are centered on lakes and islands.

Some facts:


And we are going to go there.

And we are going to deploy the boats.

On an active volcano.

And it is going to be grand.

Stay posted...

We leave for the Philippines in 5 days. Keep up-to-date on this blog!

Facts from:
*Greenland is the world's largest island that is not a continent.

Tuesday, August 23, 2011

The results of the Moon Pond Test

Whilst in Maryland we attached (using the most advanced scientific methods) a 5TE sensor to the boats. This sensor was able to collect data on electrical conductivity & temperature.

Back in the lab this data was feed into the computer which then spewed out some fantastic results.

The video below shows spacial graphs of electrical conductivity and temperature. As you can see, these readings varied quite significantly across the lake.

Some other views of the pond 

View 1

View 2: Arial view

View 3

Tuesday, August 16, 2011

Preliminary preparation for the Philippines

The CMU airboats will be deployed in flood waters where they will collect important information. The onboard camera can feed back images that will show the operator where dangers exist in the water and possibly even where survivors are.

There is a saying in Robotics: “Whenever you go to test in a new field everything breaks”.

Therefore, no matter how many times we can get the boats to work in a lake / pool in the US we can never be sure that they will work in a flood zone until we actually test them there. That is why three of us at CMU will be heading over to the Philippines at the beginning of September to see if the boats cope with a new environment. The Philippines gets great monsoonal rain every year so, chances are, we will be caught in some sort of flood whilst we are there.

World wind maps strike again

As I have previously discussed in this blog, we are using Worldwind Maps to navigate the boats. These maps, like all maps, are not without faults and have once again delivered some fantastic results. Here is our test site in the Philippines shown on Google Maps:
Victoria, Philippines shown on Google Maps

Above: Victoria, Philippines shown on Google Map

And here is the same location shown with Worldwind maps:
Above: Victoria, Philippines shown on World Wind Maps

As you can see it must have been pretty cloudy the day Mr World Wind took his photos. Hopefully we can remedy this problem before we leave.

Keep up-to-date with the Philippines on this blog. We promise a few posts whilst we are there.

Monday, August 8, 2011

Testing in Maryland, a statistical summary.

Days 2 & 3 of testing in Maryland continued without many hitches.

In fact, Day 3 went so well that three of our brightest opted to stay on for an additional day of testing... aptly named Day 4.

Below is a statistical summary of the four days of testing...

Throughout the testing:
  • We were at the pond for 41 hours.
  • The boats were in the water for 22 hours 14 minutes. They were deployed 77 times. 
  • The average deployment time was around 17 minutes. 
  • 18 feet of Subway, 2kg of Chinese and 3 large pizzas were consumed. 
  • The boats travelled approx. 26.23km.* 
  • Max temperature reached 108°F!
*this figure has been estimated because we can’t be bothered working it out..

Graphical data
  • Graph 1 - Number of deployments 
Boat 1 was used most (after the rescue raft of course)
  • Graph 2 - Average deployment time 
Average deployment time increased over the weekend as bugs were found and efficiency was increased.
  • Graph 3 - Data collection 
JoyRideMode made this possible. This graph has been added for humour and serves no other purpose.

Day 4 ended with a hurried pack-up as the Maryland’s fiercest rains rolled in. Unable to operate in the rain, we packed up and called it a day. (Yes, we are aware of the irony of having to pack up for rain when the boats are destined for monsoons in Manila. No, we don’t need you to point it out again.)

Sunday, August 7, 2011

The joys of JoyrideMode™

Moon Nurseries, where we were conducting tests, was chosen because the owners were interested in collecting data from various areas around the pond. Before we could finish the job we had to have collected data from hundreds of different points around the pond.

Disenchanted with the serious nature that the project was taking, one of our programmers decided to (secretly) implement some new code. This code has since been trademarked as JoyrideMode™.

JoyrideMode™ | joi rīd, mōd |

  •  a mode in an amphibious airboat where the boat is unknowingly being operated purely for the pleasure of the operator and no other reason. Oh damn, I had the boat in JoyrideMode all day! It didn’t log any data!
  •  Key features of JoyrideMode:
    • Disengages the sensors.

o   Battery efficiency decreases by 97%

These batteries will allow the boat to run for approximately 11 minutes.

o   Feeds out random values.

Some of the important logs that the boat produces.
o   Frequently shuts down forcing it to be rescued. 
Although they spent most of their time rescuing the boat our team remained positive
As you can see, JoyrideMode is an important feature of the Airboats.

Our chief boat operator is unaware that he is operating in JoyrideMode. For three hours he zig-zags across the pond believing that he is collecting valuable sensor data.

Here, the code of JoyrideMode is discovered for the first time. Laughs were had.

Coming tomorrow (8/9c): “Testing in Maryland, a statistical summary”

Copyright information: JoyrideMode and the 7-battery logo is a trademark of the CMU Airboat project.

Thursday, August 4, 2011

Day 1 of testing in Maryland

Today was the first day of testing the boats in a pond at the Maryland Moon Nursery and it was a great success.

We arrived in the morning at about 8am (already slightly behind schedule after almost half the group slept in).

Over the course of the day we deployed Boat 1 about 12 times and Boat 2 14 times.

Whilst in the beginning we tried retrieving the boats with fishing line. This idea was quickly abandoned and instead we borrowed a metal dinghy to rescue the trapped boats.

Yaaaay! We got it!

Boat 1

Boat 1 was very successful because it was able to go out and reach a waypoint. A sensor mounted about a foot below the boat was able to collect data on Temperature and Conductivity.

To try to get the sensor to sit below the surface of the water was quite a difficult task. The sensor had to be weighed down by something heavy enough to counteract the inertial force of the boat but not too heavy that it caused the steering to malfunction. Our engineers spent many hours searching the banks of the pond to find the perfect sized rock.

Video: Boat 1 heading towards a way point.

Image collection

Boat 1 was going so well that we decided to collect images off it. Using a fisheye lens attached to the front of the phone camera we were able to get photos like this:
Above: the photo an image from the on board camera.

Above: Here we have an image of a few local fishermen
who had become lost on the middle of the lake.
With the help of our boat we were able to lead them back to land.
They rewarded us with the finest of fresh fish 'n' chips.
Boat 2

Whilst the software and hardware on Boat 1 and Boat 2 is very similar the results were quite different. This was mainly due to the different hull size and fan size. Boat 2, although much lighter than Boat 1, was far bigger and was much more susceptible to the heavy currents in the pond. This made it very hard to keep the boat in a straight line because the powerful fan would constantly over-correct. Many hours of coding and redeploying were at no avail.

It was very exciting to see two boats in the water!

We hope to be able to launch Boat 2 again tomorrow.

Although we had a very detailed and down-to-the-minute schedule, by the end of the day the only thing that actually ran on time was our departure from the lake.

To finish this post I give you some photos of our engineers working with the boats...

Wednesday, August 3, 2011

The first "field" test

In preparation for our trip to the Philippines in September, the airboat team has decided to travel to Newark Delaware to see how the boats cope away from home.

We will be spending the next few days sending the boats around an irrigation lake to collect data. If all goes according to plan we will have three days worth of useful data that we can take back to Pittsburgh to build better boats.

The chosen test side is a closed system irrigation pond in Chesapeake City. A satellite image of the pond is shown below:

Google Map

This image, showing the 3 acre pond, has been provided by Google Maps.

Everything looks good so far? Well..

..we aren't using Google Maps for our software. Instead the current interface uses World Wind maps (provided by NASA). These maps are easy to embed and manage within the software. When we look at the same spot on the World Wind maps we see this:

World Wind map

See the problem?

Clearly, the map provided by World Wind is older. (NOTE: If the Google Maps is actually older and the pond and structures have since been demolished then we are in a lot of trouble)

Tomorrow will be interesting because, as far as the interface is concerned, we will be testing the boats in the middle of a field.