A video of Daniel McKinnon's presentation to Dronecon 2014 on what farmers really want.Read More
Last summer, Agribotix set out to discover how drones could be used to help growers make better decisions. Some of you followed the journey through our blog posts, and we are grateful for all the discussion and discourse our results generated. By the end of the summer, we had the pleasure of working with dozens of growers across many states and several countries who leveraged Agribotix drones for image processing on different types of crops.
Over the course of our first full year in operation, we found that many growers are looking for a simple, cloud-based solution to process their images into actionable intelligence. Today, we're launching our drone data processing system -- Bring Your Own Drone™ (BYOD) -- to anyone flying drones for agriculture.
The service takes images from virtually any drone, stitches them and returns a single view of a field. If you send us near-IR pictures, we'll return stitched results with a false-color NDVI image as well as a shapefile that can be imported into virtually any farm management system and used as an aid to precision fertilizer application.
If you are using a 3D Robotics flight controller, you can download our Field Extractor software, which will automate the process of selecting the images for each flight, geotagging them and uploading.
We've made the process as risk-free as possible; we process your results, return a thumbnail, and you only pay if you like what you see. You can sign up and begin processing immediately. Use the Discount Code FIRST-FIELD-FREE and your first field will be processed at no cost.
If you are flying drones for agriculture, we hope you will give our service a try and let us know what you think.
-The Agribotix Team
In our two previous GoPro blogs (Part 1 and Part 2) we've experimented with ways of hacking the GoPro to make it useful as an ag drone camera and pointed out some problems. We solved most of the problems, but Part 2 left us with a show stopper: the shutter speed of the GoPro can't be set by the user. It's not a problem in bright sun, but in low light the shutter speed gets as low as 1/60th, which is disaster for an aircraft moving at 15 m/s.
Or is it? Jimmy was out flying last week right around sunset and he returned some absolutely stunning imagery at around 1/60th. The shot below is a zoomed-in bit of an image showing a dirt road about 5m wide. Note the fence posts along the left edge that are sharp. With the plane flying at 100m AGL and 15 m/s, the camera traveled 0.25m while the shutter was open so the smearing should have been horrible. But we can't detect any at all. What happened?
We're still trying to figure it out, but we believe it is due to the rolling shutter on the GoPro's CMOS sensor (and all CMOS sensors). Normally rolling shutter is considered an annoyance (e.g. propellers that turn into scimitars), but it appears to be saving our bacon in this case. Our theory (yet to be proven) is that the rolling shutter scans through the entire image in the shutter-open time slot, but every line of pixels is exposed for only a fraction of that time (if this isn't clear, there's a nice little animation in this link). We're doing some tests on the GoPro with moving objects to try to get a more quantitative handle on it so we should have some more refined ideas soon.
So is there a downside? Perhaps. The way we orient the camera on the plane, the bottom of the image has moved forward 25cm relative to its location when the top of the image was exposed. Effectively we are compressing the aspect ratio of the picture a bit in the landscape direction. But this is less than 0.25% distortion (25cm in a 100m image size) under the worst case (normally the shutter speed is higher than 1/60, even on dark days). I expect that the inherent lens distortion is even higher than that.
The jury is still out on this theory and we welcome comments from anyone who perhaps knows more about CMOS rolling shutter issues and/or specifics of the GoPro. But tentatively we are withdrawing our "show stopper" verdict on the GoPro. Now if we could only get the cost of the unit down...
Our neighbors to the north have taken the lead in setting reasonable rules for UAV use. Transport Canada announced at a conference in Montreal yesterday that UAVs under 2 kg do not need a Special Flight Operations Certificate. Operators still need to comply with regulations very similar to our own AMA rules (i.e. a set distance from airports, AGL ceiling, etc) but there is no absolute ban of commercial operations. Along with this carrot comes a large stick in the form of steep fines for companies that don't comply. This is exactly the sort of regulation that US operators are clamoring for. To Jim Williams and the rest of the band at our FAA -- are you listening?
A few weeks back we wrote a blog on modifying GoPro's for use in drone ag imaging. If you recall from that blog, there were a couple of open issues with the GoPro:
- We couldn't set the clock accurately because there is no GPS on the GoPro.
- The images using the 2.97mm lens from Peau Productions wouldn't stitch with Agisoft Photoscan Pro.
We've solved those two problems, but have identified another that might be a show stopper for using the GoPro.
Setting the time: We found that the GoPro downloads the clock time from the iPhone or the iPad when it is connected via WiFi. Most of the time this happens automatically, but if not press the little wrench icon on the i-device, then scroll down to "Set Date and Time." Then the trick is to make sure the i-device has the correct time. In IOS, if you set the Settings/General/Date & Time to "Set Automatically" you might think that this would set the time automatically -- but you would be wrong. Unfortunately you'll need to toggle the "Set Automatically" switch on and off. To make sure it is really done correctly, we suggest that you download an atomic time app. We find that the GoPro's drift between 1 and 3 seconds per day so this is something you'd need to do every day. So -- this works but we're trying to make the work flow as simple as possible and it adds a bunch of steps. (BTW, we haven't tried it on Android devices but the process is probably similar).
Image stitching problems: We bought two 4.14mm lenses from Peau Productions -- one with the regular IR-cut filter and one "NightVision" model (i.e. no filter). Then we installed the Event38 red-notch filter as we described in the previous blog post. We're happy to report that all the "squigglies" we saw last time have gone away. Here's a zoomed-in example of our model airfield. It doesn't have the nice straight lines of row crops but you'll get the idea:
A vibration mount is pretty important for use on a multirotor. There are lots out there, but this one from DJI works well for us and isn't expensive.
So, we're happy campers, right? Actually no -- the GoPro has a HUGE problem that at the moment will keep us from using it. Namely, we can't set the shutter speed. On bright sunny days the shutter speed will be high enough, but on overcast days it drops as low as 1/60th. With our planes flying around 15m/s we really need to keep the shutter speed up around 1/1000th to obey the rule of thumb of not moving more than one half a pixel while the shutter is open. So if anyone knows how to hack the GoPro firmware, we're all ears!
In closing, we are working on another solution but it's still too early to discuss. We'll have another blog post on it in a month or two.