Matrix Alterations Test

On June 14th, 2016 we finally caught a brief window of good weather to take the Matrix out and see if our alterations cured any of the issues we previously had.

Mike Bomber was the pilot in command (PIC) and I was the pilot at the controls (PAC) and Dr. Joesph Hupy was the spotter.

Mike flew the platform manually for a while test the loiter function as we were able to identify issues while in loiter previously.  We did not see any of the previous issues during this test. We decided to create a small mission in Mission Planner to fully test platform.  We ran the mission 4 times while we adjusted a few settings improving the flight characteristic. No issues were encountered during any of these flights.

I examined the same Vibration and Max Consistency XYZ reports in Mission Planner to compare with previous good flights. The reports are not as good as previous flights but they are certainly better then the flights where the issues occurred.

(Fig. 1) Vibration report from flight after alterations were made to the Matrix platform.

(Fig. 2) Max Consistency XYZ report from the flight after alterations were made to the Matrix platform.

Additional research and flights will be conducted to continue to improve the functionality and reliability of the Matrix platform.

UAS Platform Alterations

In late May we were flying the Litchfield mine site and the Matrix platform was not following the flight plan as precisely as desired. We speculated it was being affected by electromagnetic interference (EMI) but were uncertain so we grounded the platform until we could identify the problem.

I worked with Mike Bomber who is the current UAS technician to try and locate the issue with the platform. I downloaded the Flash Logfiles from the Pixhawk to examine the various reports created. The most noticeable variation between a good flight and the flight which had issues was the amount of vibration (Fig 1 & 2).

(Fig. 1) A good flight vibration readout from Mission Planner Flash Log files.

(Fig 2) The vibration readout from the flight when incurred issues. You can see a drastic increase in vibration near the end of the flight plus error logs were recorded which sent the platform into a fail safe mode.

Secondly we analyzed the Max Consistency XYZ which displays the raw magnetic field values for x, y, and z axis. The information contained here relates to the GPS location. The graph shows there was no consistency at all during the trouble flight (Fig. 3)

(Fig. 3) Max consistency XYZ from the trouble flight. You can see issues throughout the majority of the flight.

Due to the recent issues with the Matrix and information from the flight logs we decided to move a few items on the Matrix frame to reduce vibration and eliminate possible interference between electrical components. Additionally we changed the GPS to a newer model which utilizes various satellites for better location accuracy. 

We moved the the Pixhawk from the middle of the frame to the top utilizing the mount and bubble top from the Hexacopter. The plate was mounted on rubber dampeners to help reduce the vibration to the Pixhawk. 

Additionally, we installed the new GPS with a copper grounding plate mount. We had installed the same model GPS on a different platform and we achieved connection with 20 satellites.

(Fig. 4) Matrix platform with changes made based on the information from the Mission Planner Flash Log files.

  We hope to be able to perform a few test flights in the upcoming weeks.

Litchfield Mine--03/13/2016

Introduction

Today for class and we headed out to the Litchfield Mine in Eau Claire, WI.  Our class intended in collecting GCPs for a series of flights to be flown by Peter Menet of Menet Aero.  The objective of the flights was to calculate new stock piles of various aggregate piles from the mine site (Fig. 1).

(Fig. 1) Aggregate piles within the Litchfield Mine Site.

Due to an unforeseen issue with the GPS we intended to collect the GCPs with we were unable to gather any GCPs for the site.  In the future our class will be exploring calibrating these images with previous images which were captured with GCPs to see if we can obtain the same accuracy without collecting GCPs every flight.

Methods

The flights were conducted by Menet utilizing his DJI hexacopter (Fig 2).  The hexacopter was rigged with Sony ILCE 6000 digital camera (Fig 3).

(Fig. 2) DJI hexacopter owned by Menet Aero.

(Fig. 3) Sony ILCE 6000 rigged on the DJI hexacopter.

Menet flew 3 different missions to assess the results of flights with various heights and quality of images. Menet flew the flights with the following parameters.

• 200 ft elevation and 12 megapixel resolution
• 200 ft elevation and 24 megapixel resolution
• 400 ft elevation and 24 megapixel resolution

The missions were created utilizing a mission planner software created by DJI (Fig 4).  The DJI software is very similar to the Mission Planner software which I have utilized in past blog post.

After all of the flights were conducted I input the collected data in to Pix4D and created an orthomosaic image for each of the flights.

Results

( Fig. 4) Zoomed in image of the results from the 12 MP and 200 Foot elevation flight.

(Fig. 5) Zoomed in image of the results from the 24 MP and 400 foot elevation flight.

(Fig. 6) Zoomed in image of the results from the 24 MP and 200 foot elevation flight.

Additionally, I wanted to compare the results of volumetrics of a stock pile between the images. I utilized the volume tool in Pix4D to calculate the volume (Fig. 8).

(Fig. 8) Way points from Pix4D to calculate volume from.

(Fig. 9) Display of the volumes taken from all three images. 

Discussion

From examining Fig. 5-7, I feel the 24MP image collected at a 200 foot elevation had the best image quality.  The 12MP image collected at a 200 foot elevation had the second best image quality.  The 24 MP image collected at a 400 foot elevation had the worst image quality of the three. While image quality is one aspect I was examining, I am also taking in to concideration the amount of time it takes to process the image.  The initial processing times in Pix4D are as follows:

• 24MP (200 ft) : 1 hour 25 minutes and 15 seconds
• 24MP (400 ft) : 32 minutes and 21 seconds
• 12MP (200 ft) : 39 minutes and 2 seconds

I was hoping to keep track of the full processing time between all three of the images.  Unfortunately, my schedule did not allow me to babysit the computer to track the full processing time of any of the images. So my judgments will be based off the above processing times.  I feel the 12MP @ 200 ft has the ability to be the go to set up depending on the application.  Obliviously, the 24MP @ 200 ft offers a better resolution. Most applications will not require the resolution of the 24MP for the desired results of the project.

Conclusion

You will need to consider the desired out come for your project when deciding on what the quality of your imagery should be.  Each sensor will result in different outcomes.  Testing your specific platform and sensor will give you the best knowledge for selecting the best parameters. 

First Nice Day to Fly--3/7/2016

With abnormally high temperatures for early March Dr. Hupy decided to take advantage and head out in to the field. Mike Bomber and Dr. Hupy made a number of upgrades to multiple platforms over the winter which require calibration and testing.

Dr. Hupy decided to meet at South Middle School in Eau Claire, Wisconsin around 11:00 am. South Middle School is where the majority of our flights take place due to the accessibility, location, and various features to capture with the imagery. Dr. Joseph Hupy, Mike Bomber, David Leifer, Mattheus de Waard, and myself made up the group for the day.

The first objective for the day was to calibrate the Hexacopter (Hexa).  Last year, the Hexa had a broken wire causing malfunction with the flight controller resulting in a crash. Multiple upgrades have been made to the Hexa since the crash, including a new GPS, and a upgraded flight controller. The GPS needs to be calibrated before any flights can be made.  The calibration of the GPS involves the technician to complete multiple circles while holding and rotating the platform on all axis.

The second objective for the day was to fly the Phantom which was also recently upgrade by the factory.  We experimented with multiple mission planning software programs including:

• Maps Made Easy (IPhone)
• Capture (Pix4D-Android Tablet)
• DJI Go (Android Phone)

I application platforms in the ( ) are what we used for our process. Some of the apps are available on both Apple and Android.

All of the apps worked very well and were user friendly. We flew the an area over the running track at South Middle School in Eau Claire, WI.

(Picture 1) Dr. J. Hupy (left) and Mr. Bomber (right) preparing the their respective platforms for flight. 

Results

Discussion

Looking at the mosaic image you can see some distortion with the lines on the track.  We did not tinker with various altitudes or spacing for these flights.  Checking the functionality of the various programs was the main objective. I feel these flights were successful and warrant further testing of these programs in the future.

Goals and Background

Welcome to my Unmanned Aerial Systems (UAS) journal.  I am currently a student at University Wisconsin Eau Claire pursuing a degree in Geography and Environmental Science.  While I am currently enrolled in a Geography 390: Unmanned Aerial Systems, I have also been hired as an intern/research assistant to work with Dr. Joseph Hupy on his Regent Scholar research.  Along with the research I am also preparing to take over as the UAS technician for Dr. Hupy.

I am creating this blog to document my work with UAS platforms and processing software.  To see my work from UAS class free to visit my separate blog for the class.