Method For a Robust Camera System

By teaching the camera multiple models of the same object in different lighting conditions, a change in lighting will not effect its ability to recognize an object. The more cases it it taught, the less of a chance lighting will cause any effect. Frame refresh rate is still extremely low, somewhere near 0.5 Hz. I have attempted to optimize it by limiting the pre-processes (ie. filters) the camera uses, but this has only had minor effects.

Object locate function

It appears that the cameras have an object locate function. They don't react to changes in light, and are pretty much the best thing ever created! Simply tell it what object to find, and it does, about 90% of the time!

A Little Motivation!

Resident Camera Guru

Small success with camera

Playing around with a few of the onboard sensors, a combination of useable and somewhat robust sensors were just found. Utilizing the onboard color relearning function (allows for multiple colors to be stored and the most dominate one is passed to other sensors for processing) and auto gain and exposure adjusting, the ball can easily be recognized in both normal and semi-low lighting conditions. The biggest issue is that the auto adjusting features, which without them this process would not work, cause a very slow response time in changes in motion. To correct this, the auto-iris side project must be completed.

Compass Evaluation

After inspection and evaluation of the compass we think that the compass should be similarly mounted as last year. We will need to do proper calibration which will hopefully fix the drifting gyro correction that was experienced in the past. More updates to follow.

Gantt Chart Updated

Here is the updated Gantt Chart. We are doing all right for the beginning of the semester but are behind in a few key areas that we need to catch up on.

To download the updated Gantt Chart click here
To download a free .mpp viewer click here

New direction for cameras

As a result of dissapointing real world testing Wed. a search for a new method of lighting control was started. Two auto-iris lenses were found in our "bag of tricks" and a quest to fit them onto the current system has begun. Currently the connections between the lenses and the DVT cameras do not match, however a seperate subsystem is in the works. A variety of power control chips and a PIC processor are going to be impliment to create a system independant of the main computer that will run from values the DVT will output. High hopes for the usefullness of these lenses ability to control the ever present lighting issue.

Camera Boxes tested

The new camera boxes have been tested today. They were set at a 16' depth for 2.5 hours using a 5 pound weight in the NARC test pool. No leaks! Perfectly water tight as of now, next step is to check with SEACONs attatched. Not expecting any problems.

This Week's Progress

Our frame is almost done and it's going to completed very soon.
We need to work on picking out a team name in the near future...hopefully something that's not too embarrassing.
The vision system is being tweaked but at this week's meeting we were able to find a buoy in a lighted and dark room with a light but we want to work on some of the line tracking and underwater factors this week.

Torpedo Idea

We are starting to brainstorm possible ideas for the "Torpedo device." Here is an idea that Sayre came up with as a possible solution. This might be overly complicating the idea.

2009 AUVSI Rules

The preliminary 2009 AUVSI Competition Rules are up which is leading us to start designing the mission devices and homing in on what our vehicle needs to be programed for.

You can find the rules here

AUV 2009 Design Video

This is the video that was shown to the board of visitors and used for our design presentation. It shows our current design for the 2009 vehicle.

First Meeting of the Semester

The team met for the first time this semester last night. We mostly discussed the progress we have made on the frame, which should be complete by next week, and the layout of the electronics inside the pelican case.

After being plauged by errors while trying to get last year's vehicle up and running Cusch discovered that the computer is fine but believes the problem is in the thrusters' power board. This is good news and means that we should be able to start using the computer to work on code.

The link included with this post is one that was forward to me by Cush on a quality underwater camera. Because of budget and our current design we will be using last year's cameras but might upgrade before the competition if the funding comes through.
Sea Viewer