Just wanted to share the launch of the StitchCAM on Kickstarter. The StitchCAM is a revolutionary new multispectral camera drone that simultaneously provides full color images and NIR data. Now every farmer/greens keeper/developer can have NDVI on tap at the push of a button. View StitchCAM Kickstarter for more information
Much thanks to everyone here and to the DroidPlanner and Pixhawk/PX4 teams, without all your hard work this would not have been possible.
Comments
Great comments on this blog post, now listed on the 'UAS for agriculture' and 'Aerial mapping' pages.
Phil,
A lot of our development work has been done in Iowa so we know that area well. In addition, we are closely tied to the remote sensing industry in San Diego and we agree with everything you said about this being a very challenging problem (and like your analogy about NDVI as the doctor).
We will be doing demos across the Midwest next week. Where are you located? We would love to come by and fly a demo for you also.
Jeremy - thanks for the interest. You hit the nail on the head that comparable sensors do cost more than our complete system (and our will cost considerably more after our Kickstarter ends). Our key is our patent pending sensor. At the core we are sensor designers and have spent considerable time developing our sensor solution. We use Public Lab as an example because they do a great job explaining the technology at a high level, but we are not using any of their hardware or suggestions. Like everything in the optics world, many tweaks need to be made for each application. We use specific flight planning, image collection, and processing parameters for each application. We have approached the problem as the quadcopter is really part of the sensor itself because there are just too many fundamental limitations to strapping a commercial imager to the bottom of a craft that wasn't design specifically for that imager (sensor type, pixel pitch, integration timing, filter designs, lens designs, vibration isolation scheme, thermal management, electrical noise, etc.).
You seem to have done your homework and we would love to chat more. Are you in Cali? We would love to come by and fly a demo, the product speaks for itself.
@Jeremy - WOW! Your comments are a breath of fresh air. First off in full disclosure, I am a partial owner of RoboFlight. Our primary delivery is remote sensing data analysis, with a heavy focus in Ag. We also manufacture UAVs (have a new platform out shortly that is co-developed with Rite Wing, the creators of the Zephy and Zephyr 2 - running on the fabulous PixHawk of course) as well as offer a robust UAS flight school that teaches end users how to properly use Mission Planner for both copters and wings as well as collect their own data correctly. From there we can process their data. We have manned aircraft in operation as well.
Our real strength is our team and their extensive background in remote sensing, particularly in processing imagery from UAVs. We are approached weekly by large ag companies attempting to DIY their own fields, only to find the data incorrect. We have looked at their data sets and they are a mess. Most are using copters or planes they built themselves and then using off the shelf NIR cameras and software solutions. Additionally, farmers are being sold pretty NDVI pictures that also are not "actionable". Surrounding myself with agronomy PHD's and scientists, my eyes have been opened to DIY remote sensing. It is NOT for the faint or heart, nor for anyone that doesn't have extensive background in processing and analyzing the data. There are only a few legitimate companies in the US that have such individuals on staff. I hope SNAP does as well but I can't find any information about their team online, so I wouldn't immediately trust anyones camera/sensor package nor software. I'm VERY skeptical of any software that can do a proper field ortho creation and evaluation in 10 minutes, but I hope to be proven wrong.
That SNAP camera is a very broad NM band for calculating NDVI. I see your comment about calculating resolution dependent on megapixels, which is why I asked what the megapixels were, but I did not get an answer. The Sensfly camera is only 2 megapixels, which makes processing faster due to less data, but you are limited on the high resolution needed for doing early season evaluations of most row crop. Also, the video implies you can use NDVI to detect drought spots. That isn't actually true. NDVI is only indicating stress levels, not the specific stressing cause such as drought. It is better than nothing however, especially compared to what is seen with the naked eye. A metaphor would be this: NDVI is like a doctor taking a patient's temperature. It can tell that the person is sick, but it will not diagnose the disease. Same with NDVI.
Also, farmers need a unit that can cover 200-500 acres in one flight. There is certainly a place for rotors, but not on large farms. They don't want to land every 40 acres, charge batteries, etc. It is a waste of their time and is inefficient for farmers. Second, the farmers that will create a tipping point of adopting UAV remote sensing imagery are in the midwest, and are growing corn, soybean, etc. These are the farmers who's smallest field is 125 acres. In states like Iowa, Illinois, Nebraska, Minnesota, North Dakota, South Dakota, Indiana, Wisconsin, Kansas, Oklahoma, and even Texas, the winds during the growing season will eat a rotor alive. A n aircraft must be able to fly in 25 mph without concern.
If there is one disturbing matter in this entire excitement of using UAV's in precision ag, There are countless UAV companies that keep popping up throwing out terms like "NDVI" and most of what they know is from what they read on the internet. Once an agronomist writes a prescription based on bad NDVI data and it causes damage to the field, I think we can see farmer's adoption rates of the technology go down as well as law suits go up. You summed it up best.....ask ALOT of questions on both hardware and software.
At that price point I also question first and foremost the ability to capture accurate data with both....RGB and NIR filtering.
How is it done?
Let me be clear that I'm not here to bash SNAP, I welcome all to our community as it would apply to advancing our position with the FAA.
Back to SNAP....anyone who has done their research/homework on this topic knows: Any "sensor"/camera(just it) worth their weight will cost as low as half - to double/triple the "kickstarter" pricing for the complete package.
I for one will not endorse a $130 camera advertised on P*^L%b to mount on my UAV to collect data in the Ag business. My client the (grower, PCA, scout,agronomist) will not stand for data results less than accurate. This is not a game to them, especially in Cali. where the drought conditions dictate everything.
Give me the "steak" not the "sizzle".
Everyone, please ask lots and lots of questions, don't assume anything.
Also, his answer about the 40 acre flight is about right for a multirotor tracking along at about 8 fps with no "loitering" along the way between waypoints, flying about a 10 min mission. (typical results based on conditions, i.e. temp, winds, etc). Is he flying a "vertical" or "oblique" camera angle as well?
Phil, I'm curious to the answers of your questions, especially as it would apply to the camera "sensor" question. His response seemed somewhat vague to me. The answer is pretty simple. How many acres covered at what resolution (megapixel), determines altitude flown based on FOV without distortion, 15, 30, 60 % overlap? detemines flight time/aloft, etc.
To me it's all about the "sensor" first, platform can take a backseat. In the final analysis is all about the "interpretation" of the data. The is no set "standard" in place and your understanding of the processed data can be different than mine.
What SNAP has to offer is not ground breaking, just repackaged goods. The same came be done today with off the shelf items.
I can send you a ready to go setup today!
The arming safety feature has been moved to a combination of physical hardware and app features.
but how no radio? dont ya need a radio to arm?
-
1
-
2
of 2 Next