Utilizing Pix4D in forensics

PIX4Dcapture
1

In using for accident reconstruction the limit of a 35 foot minimum operating height does not always allow the use of the 3D orbit function to capture vehicles since most vehicles post collision are in wrecker or auction yards with multiple vehicle sitting very close to them.  The height of 35 feet is sometimes too high.

 

Also and very importantly, in flying these same vehicles you cannot see them in the overhead view on the app that is used to set up your grid to fly, since this view is a google earth type image.If the grid could be edited or seen when flying above an area with the live view mode in the upper right corner, this would help greatly.  Right now it takes a ton of time to try to get it to work using estimations.

2

Hi Jason,

It is always nice to have feedback from real cases and see what could be the pain points for users. I have shared your interest in the possibility to go below 35 feet with the developer’s team. If not confidential, is it possible to share pictures/projects of the kind of scene you would like to reconstruct? Is there buildings around?

Note that you can fly a Freeflight mission lower than 10 meters above ground level with Pix4Dcapture, however, it requires some skills and I recommend to use it only if the other features in our app does not give you satisfying results. You can learn more about planning and flying a Free Flight mission at How to prepare and start a free flight mission. 

For the second part of your feedbacks, if my understanding is correct, it is difficult for you to center properly your circular missions flight plan above the object you would like to reconstruct? Is your reconstruction impacted? Have you considered flying other types of missions? (double grid with a tilted angle for the camera for example)

3

Sorry to bump an old post, but I just wanted to echo Jason’s points, specifically the one regarding setting a grid pattern only using Google Maps. At times, it is extremely difficult to see how much area you want to capture at a crash scene, especially if the google map is in an area of poor quality, no cell service, or a section of roadway with minimal identifying features.

What I would love to be able to do is manually fly a perimeter, manually setting points along that perimeter while flying, and then have a grid pattern built from my perimeter flight points. This would give me a much higher degree of confidence that I am capturing only what I need, avoid problem areas (like trees) or build an appropriate grid along a roadway curve.

Right now, I try to walk the perimeter with the controller, as it shows the controller position in the map to find my boundary, and set it that way. Far from precise, and I don’t always get it right, and it also really slows down my scene processing times, to the point I usually just end up manually flying it.

Thanks!

Justin

4

Hi Justin,

We encourage all users to share their feedback and it is always valuable to know what could be improved in our application to match the best as possible with our user’s needs.

In the future, we will go more into dedicated applications for different fields like forensics. Then I shared your feedback with the developer’s team so that they take it into account.

Best,

5

Justin nailed it. If there was a way to setup a perimeter while flying utilizing the Pix4D App and marking the perimeter by using the camera capture button on the remote for instance this would be a game changer.  The 3D grid tool could then generate just inside this perimeter.  The perimeter being utilized of course as the maximum length and width of the grid.  Too many times have I set up a grid and had to manually take over the flight to keep the drone from striking a tree.  This cause me to start over with a new grid placement.  If I can fly a grid perimeter and set maximum points as far as width of the grid then a can avoid these objects.  The same would work with cars in wrecker yards.  Right now I have setup the 3d orbits using the position of the drone and the remote on the screen.  This is an OK work around with vehicles however with long scenes on the roadway this is much more difficult.  Also the height restriction is tough on some roads with low power lines that cross the roadway at an angle near the collision site with outstretched tree limbs directly above them.  I have to resort to in these type situations flying the drone through the center of the scene and then down each side of the scene with the drone camera perpendicular to the roadway all while videoing these three runs then pulling the frames out of the videos with Pix4D desktop.  This is a workaround for now with the height limits.  The main point is having to start over multiple times drains batteries, which if you run out of batteries you have to charge.  Charging takes time.  I keep 4 batteries charged when I go into the field but can deplete all on large scene before any are recharged.  This causes unnecessary time sitting on the side of the roadway.  The more we can lessen roadway time the safer we all are.  I am a private reconstructionist no longer in Law Enforcement.  I no longer have the luxury of lights that the motoring public pay attention to and can no longer shut down roadways, so the less time on the side of busy roadways the better.  

6

Hey Jason, if you’re still around would you mind emailing me at max.davis@dps.state.nv.us? I’d like to ask you some questions about flight planning and data acquisition for accident recon. I’m an investigator with one of the major crash units within the Nevada Highway Patrol and we are in the midst of testing Pix4D on some of our crash scenes.

7

First, as a reconstructionist, I love Justin’s suggestion and that would be extremely useful for crash scenes. 

Jason:  I also use Litche for vehicle orbits.  I believe it allows you to go as low as 16 ft.  I’ve also set the drone to record video and hand held it while walking around the vehicle.  This is useful for crush detail. Video isnt as high quality as photos, so if you have someone to manually click off pics as you’re walking around holding the drone, that’s even better.

As far as crashing into a tree, I got a tip from a Delaware Helicopter pilot.  They perform what he called " MOCA", Maximum Obstacle Clearance Altitude. Basically, when I first arrive on scene, I manually send the drone up with the camera parallel to the ground; facing the horizon.  I continue to climb until the highest structure (tree, poles, signs etc) are just below that horizon.  I then rotate in place 360 degrees making sure everything in the immediate area around my scene also rest below.  I then add 10 ft as a buffer to the altitude and set that as my flight height in the Capture App. I can the map my scene without worrying about colliding with anything in the area to be mapped.  I have a MS Word file with screen shots illustrating the process if anyone is interested.

On a feature side note outside of crash recon, It also boggles me how Pix4D hasn’t included an address search field, so that the user can plan flights prior to being on scene.