# When Precision Agriculture Meets Autonomous Drones

- Event: NestGen ’22
- Published: March 4, 2022
- Length: 49:31
- Watch: https://nestgen.org/library/watch/IuUWtAZ1tec/

> The transcript of a recorded NestGen session, for reading or for use with an AI tool. It comes from the video's
> captions; automatic captions can mishear names and terms.

## About this session

Find out how autonomous drones are helping make crop yields more profitable and sustainable through systematic data collection & processing.

## Transcript

**[0:12]** hello everyone and welcome to one of the most exciting sessions of NestGen 2022 with this session we will find out how autonomous stones are helping make crop wheels more profitable and sustainable through systematic data collection and processing I'll start with the introduction to our speaker today today we have darren from agribox darren is the founder of agribox a drone service provide provider company with a focus on agriculture research he attended florida state university and graduated with a bachelor's in information technology he earlier worked for a startup in the net in the network service and management industry for five years it transitioned from providing services to small businesses to managing network devices for the depart for department of florida state university daryn currently spends his time helping the agriculture community understand and implement drones into their current operations sideways has partnered with agribugs to closely work together on on the advancement of technologies to

**[1:14]** support autonomous drone operations agribugs has been a pioneer in field of precision agriculture across the us constantly contributing in community outreach and collaborating with the university of florida with this I hand hand it over to you darren hey thank you uh for uh handing it over to me I thank you everyone for joining us here at the nextgen 2022 um conference um in regards to precision agriculture meets autonomous drones so we're going to start basically agrobugs is a drone service provider I provide multi-spectral and uh color imaging for research staff and commercial farmers throughout the area in the southeast united states here in florida and south georgia but before we go into the full autonomous drones in regards to agriculture and how is it affecting and how it's going to help the industry I would like to go through a little bit of the data and how the data kind of stairs um the pathway to autonomous drones one of the payloads we currently use

**[2:18]** um is a DJI x5 camera to give us some rgb uh color images um rgb images is very unique in regards to giving us a good color basis of what the field in agriculture is looking like for the client for the farmer this helps because there's no translation and information to the farmer when he sees a photogrammetry of an imagery of color images he kind of sees the anomalies for himself so there's no translations or no different colors there's no um information that kind of uh needs to be translated to the farmer but he can actually see the data he can see anomalies for himself and he can help ground truth the data I'm speaking regards to ground truthing is very important ground truthing is uh when you can take the data that we basically create and go out into the field and look and see what those anomalies are in the actual information that we're providing

**[3:21]** the layer that's most important for agriculture and for the client in this industry is I guess the word trust trusting the data and the farmers and the clients need to trust the data so ground truthing this information uh creates that level of trust that can translate to the actual farmer this kind of level helps the farmer have trust in your data and that there's some type of value and information that you're providing to the client so this uh image on the left is a citrus field it's 40 acres so one of 16 40 acre plots from this rgb imagery we can basically extract these levels of data elevation and a false color ndvi some may know in industry that this data some has been processed by drone deploy um and this is the level of maps you can get from just an rgb image uh and like I said this kind of data provides uh a one layer of trust to the farmer to

**[4:27]** know what's going on in the field for example the elevation information from that that rgb image can say there's spots in the rgb image that there's missing canopies that they can actually go out into those areas and ground truth that information so you can create that layer of trust from the beginning with the farmer to say hey this is what's going on in the field this is anomaly and this is what we're picking up as rgb imaging and agriculture has not just started with drones it started with aerial imagery from cessna planes from satellites but this layer of high resolution data has not been presented to farmers so they can see more information from this layer and providing them with quality data can definitely help um gaining that trust with your client and basically help move the progression on to getting more business from a farmer and continuing serving them in the industry

**[5:30]** another uh payload we use and what I focused on is multi-spectral imaging multispectral we basically captured data in a different wavelengths uh that the visible eye can't see um we captured in the visible light non-visible light and infrared thermal and for people in the industry and agriculture uh most people should know about ndvi thermal and uh uh these nd re uh kind of images when I first started we started with the slant range system which is a kind of a one a stop shop where a sensor is basically paired with a software so we can go out in the field with that slant rain soft uh sensor go out in the field fly a field and basically process the data in the field and basically provide in uh insights within real time on the field to a farmer or to a crop consultant to do some extraction of data or do some ground truthing within a

**[6:34]** couple hours of actual flight time so uh one general um trend what I'll be speaking about is time uh time from data collection to time to take an action is very important in this industry and the issue what we have been seeing is uh heavy data takes heavy processing time and the more heavy the data data is the more the length of time it takes to actually create an action in the field um and since we are dealing with plants data and environment stuff is changing within days diseases can spread within days and trends so the faster we can actually get data out to a farmer to a crop consultant is better for uh serving or making a decision in the field another sensor I use is the micasense autumn this basically sensor is paired with multi-spectral um imager imagers and a flare

**[7:34]** thermal imager which is very interesting because now we can get some good thermal imaging um before we use flares we weren't able to kind of create photogrammetry maps um but now in with the microsets autumn it's kind of been worked with flair to kind of join that operation together we're now we're creating these maps uh with the thermal maps so you can see in the middle image there's a thermal map that I created um this was an irrigation research now done with a big irrigation company where we wanted to see the time of saturation in the field um to the amount of time it takes for uh the field to absorb that type of irrigation and kind of uh set the next time for the next irrigating in that point so you can see that that triangle here that purple area it was recently irrigated um and other areas haven't been uh irrigated but that's some maps that couldn't have been done in the past with thermal imaging um

**[8:39]** the layer the image to the right of that is the ndvi image uh this was an area of research where we did uh where we did not irrigate a certain area of the crop and we wanted to see it stressed um so they didn't tell me exactly where this was this was on a corn uh uh location about 150-175 acre pivot um and you can't see this from the ground so after I flew the actual drone we capture the data we process the data you can see precisely of where there was a stressed area in the location so ndvi which is normal normalized difference vegetation index helps provide that type of information to the grower into insights to create some type of tasks that they can do elevation maps are also definitely helpful because what we've seen is that in lower areas in higher areas uh irrigation

**[9:40]** uh spraying kind of can get trapped in certain location that can cause stress so if you have plants in a area that is low that's been a rainstorm has came by that area those plants might get more stress where they might get more water and can show different levels of information in that area with elevation maps and that's what farmers kind of look at in regards to elevation to see what's going on in those areas so they can kind of uh not irrigate that area or over irrigate that area just to kind of keep their uh farm to a very uh level basis so these are ag data processing softwares um I kind of worked with mostly all these softwares uh at the beginning of processing data my rgb data that you've seen in earlier slides is with drone deploy which can take your rgb color images and give you some three levels of

**[10:41]** layers of data analytics for the rgb images so drone deploy is an easy to use user-based uh software uh pix4d agisoft solvi and web odm are very good software the pix4d is a little user friendly I currently process in agisoft web odm and agisoft help process help the automation process and means it has command line so you can kind of automate some information that you want to do with processing data so processing data is kind of a bulky process in regards to you have to remove your data from your drone or from your sensor and kind of basically process the data manually meaning uploading the data to these softwares and then processing the data in regards to reflection panel uh information uh and takes these things could take hours to actually uh finish processing data so if you

**[11:43]** have multi-spectrum images that are you know 20 30 40 gigs of data that you're running through this software it takes hours to process this data and like I mentioned before the time to get from capturing the data to the time of delivering the data to a crop consultant or to a a farmer is definitely uh valuable one of the interesting uh software's I also use was solving it's always a good ag data platform um they uh definitely process a lot of my multi-special data for me the issue is is uh solvi is a data collector in regards to uploading information so in regards to automation in the future these softwares have to kind of look at how they can have a ground-based software where data is fed directly into them after a flight is done where it's not transmitted into uh transmitted into the cloud

**[12:44]** because that can take hours based on where uh the software is located or where the actual farm is located um when you're out in agriculture usually the internet or the cell tower is really you're working on 3g 4g uh kind of data set so you're not uploading you're not able to upload as fast as you can as being at on a regular um uh on a regular home on a regular medium a better high speed view um so capturing data and feeding it to a software processing on-site is something that all these softwares are going to have to be looking at in the future and command line for automation where you can feed it automatically instead of manually taking and removing data from a sensor and feeding it to a software for processing cutting that time would definitely be something in regards to automation

**[13:45]** and what we have to look for in the future when we create a drone in a box on a field so decisions and drone data um I think is uh very important in regards to the resolution and the quality of the data so when you feed one of these ag software processing programs data you have to make sure that the data and the qual the quality of the data is premium um because what you feed the data what you feed the software is what you're going to get out on the other end so if you feed software these processing softwares with images that are unfocused that have clouds in them you're going to get anomalies in your actual data sets that's going to cause false pretenses and false issues in the field when a crop consultant or a farmer actually

**[14:45]** ground treats that data so and the reason why they have to do this is because they don't trust the data based on the quality of the data that's getting received into the software so when you're collecting rgb or multispectral data in in agriculture you need to make sure that the data is very good very premium data and that is specifically in regards to when in time to capture the data it's capturing the data between 10 and 2 o'clock is very uh is an actual um a standard for agriculture data based on cloud's shadow so capturing premium data is also is very important for pushing into this agricultural ag processing software so you can get good results so you can go on a farm and translate those good results into good ground truth and when you basically get good ground truth those those that information turns into

**[15:50]** tasks for the farmer so management decisions um these decisions are based on that type of data so if the ground truth of an area that is stressed with disease is shown on your uh multispectral data or your rgb data you can basically the farmer can say hey I would want to spray this now that turns into an action uh the main goal of automation in agriculture is turning the data into some type of action for the farmer and specifically an action that saves the farm from uh uh any type of extra stress type of uh and uh mostly uh input about the in regards to fertilizer uh any type of fungicide uh if you can spray a smaller area it's better to be more focused so I'll tell you a little bit about some research uh that I've done in 2015

**[16:52]** uh when we started with multispectral cameras this is with dr matthews barrett here at the university of florida uh we did work with gummy stem blight in regards to a watermelon field so the purpose of this is that a crop consultant said that hey there's an issue happening in the field they didn't tell us exactly where it was so we went out with the drone and we basically threw the field and we looked at four or five blocks of information that is going on and four or five blocks of anomalies that we didn't know was going on so we captured these five blocks and we looked and we processed the data and we saw these areas of interest we looked at the we talked to the crop consultant and the crop consultant went out there and confirmed that this is gummy stem blight a very uh a very large disease in watermelons um what we did now is once we identify the disease we kind of map the location for one

**[17:52]** month every week at the same time uh mostly on that same day and what we did is we kind of mapped the disease and the disease spread we went from mapping one block to the spread moving into three to four to five blocks and with uh the data we're able to see the spread of the actual disease and relaying this information to the farmer he was able to spray these blocks and two or three outlying blocks that helped him kind of save the rest of the field so this is 60 acres and he was able to spray about 10 acres of this field and which basically helped them reduce costs in uh input to the field and it helped them save the amount of time it took them to spray the whole field which is very important for the industry in regards to autonomous drones this is the basis of autonomous

**[18:55]** data meaning finding this type of output finding it autonomously relaying this information to a farmer to a crop consultant to trigger that type of task to move forward and help the farm uh be proactive in solving the issue so data competition in regards to drone uh information uh maps so I remember sitting uh five years ago with a group of researchers um and they stated to me that drones were not going to be in agriculture for a long time because satellite data was going to be the competition um and at this time um let me express you the resolution of satellite data here so one of the on the left is satellite uh landsat 8 ndvi images at 30 meters resolution per pixel as you can see it's very pixelized but you're talking about a satellite uh

**[19:58]** providing that type of data which is a low-cost data that farmers do use at this time um planet ndvi is giving out three meters as you see it kind of improved the issue with this type of pixels and this type of information is that you cannot create tasks automated tasks from this type of data this type of data needs tasks that uh this type of information I'll start excuse me this type of information is not in regards to quality you kind of create a trigger for robotics to start spraying so if you see the drone data on the right at five centimeters this is basically something that I uh this is a map that I created uh alongside of this to see the actual resolution of the actual data and to get that type of precision in regards of drone data so the goal for me of presenting is that we have to kind of trigger task

**[21:01]** for someone on the ground agriculture equipment nowadays is kind of all technology based every tractor every sprayer has some type of gps component has some type of technology that is on that on that tool right now and drone data kind of is going to push the information to these other robotics on the field to kind of spray precisely when you're taking data from a satellite it's kind of hard to relay that information into precision on the field when we can get drone data now to precise to get it to the field we can basically create tasks for on-site tractors for on-site sprayers for farmers for ground shoes for crop stores it's kind of triggering multiple things in the field and this type of resolution you could be more accurate than satellite data at this time and yes

**[22:06]** satellite drone data right now at this time is more expensive then satellite data because there's manual labor in uh intense into creating this type of data and this is what I'll be moving from moving towards here um so collecting and talking about manual uh manual processes of collecting the data and looking at time so for data collecting processes you have to actually in regards to agriculture you have to travel to the farm you have and farms are not located near uh you know a couple miles away usually I travel you know 30 minutes to an hour three hours away to a field to map a a location uh when you get to the location sometimes you don't know what's going on with the weather at that location since this is where in florida and the weather changes there uh normally all the time and in regards to clouds it's another issue

**[23:08]** that discussed in the future but you have the time to travel to the farm set up your equipment collect your data and fly pack up your information pack up your data actually ground truth your own data or check your data uh on site to make sure that you don't have to go back to the data back to the field to basically fly again um then once you return to base you have to upload your data for processing now this takes a couple hours if you're processing on-site or remotely so when you process on-site this might be a little bit quicker um in regards to like I said I use the software called agisoft what kind of process on-site but which still takes hours to process um then you want once you get a finished product of processing you have to package that up and deliver it to a prop consultant to a researcher to a farmer for them to analyze the data for them to look at the data and kind of create some type of results some type of action now

**[24:13]** this person is crop consultant or farmer or going into the field and ground truthing that data creating that level of trust so whatever tasks they do based on this data is not false so going into the field and ground treating the data and making sure what our drone data is saying is true is another time so this time between getting your uh um trigger to actually go capture the data to actually creating a task can take between two three four days and that's as fast as uh as you can in regards to getting the data to a crop consultant and creating a task so this is where automation and the future of autonomous drones are gonna uh kind of take uh take the point of replacing this type of procedures and when replacing these type of procedures you're increasing your um you're kind of increasing the

**[25:16]** productivity and being way more proactive in getting to that task to solving that issue on the field and that would help and that would be the base basis of autonomous drones in agriculture is improving on that time so when you're improving time you're also improving costs so there's a lot of truth in regards to time is money in agriculture so if we can autonomously capture data process the data and feed it to a crop consultant within a day without manually capturing the data you have already removed the cost of fuel the cost of time to travel the cost of data collection and having a person actually viewing keeping an eye or a visual observer on the drone this actually helps an enormous way because you're talking about one full date of two full days of just

**[26:22]** capturing data and processing data so now an autonomous joe now can capture data related information to a crop consultant within a day or two you can have some kind of action and cut that time directly in half and farmers and in regards to dealing with a farm that kind of changes daily uh that helps with uh being proactive in the field um another issue is data management collecting this type of amount of data causes you know an issue with collecting uh storing data sometimes the raw data is not necessarily needed after a data is processed and after data is crown tree then after a task has been completed that data the raw data might not be needed so data management is one very important um [Music] one of the good uh ways of automating these type of processes

**[27:26]** training the software for anomalies that's this is very important because we can collect the data we can process the data but we need somebody to analyze data so right now the data is being analyzed manually by someone like me by a researcher by a crop consultant to look for anomalies this data can now slowly be trained uh by ai to look for the same anomalies that we um that we are looking for and with these saying this kind of uh moves into automation where if we can tag an anomaly to a task if we can tag an irrigation link to a task we can basically speed up where some farmer or someone on the field can actually initiate this task and solve that problem for example we did a um back in 2015 also we did a research on a watermelon field that had multiple

**[28:28]** irrigation leaks um they have a manual process of walking the field and looking for irrigation so you have it's 90 degree weather you have 50 acres of watermelons and you have one guy driving around or walking around looking for irrigation links it's not 100 uh where he's going to actually find that leak so what we did is we came out with him and paired him with the drone we took the flare camera we looked for large areas of cold spots in the field we went to those cold spots in the field and we found that there are large puddles of water hanging out where irrigation lines are cut what this did is it helped him improve the time to fix it it helped them reduce the amount of manual labor in finding uh these areas of interest and also what it does in the future for the crop is it creates a healthier uh

**[29:32]** atmosphere for the crop environment for the crops um well like I know that water kind of when it sits it can create diseases that can start in the plants and kind of spread to larger areas and having standing water in any field is no good so tagging a flare or a trigger for irrigation or some kind of anomaly to a task for someone to look at and to solve is definitely uh one of the major components of the automation processes so I guess is the big uh uh holy grail is uh bv loss uh certification and get this going and uh being legally being able to legally fly these drones autonomously um how I kind of group it in my industry or to how I look at it I look at semi-autonomous flight versus fully autonomous flights

**[30:34]** semi-autonomous meaning that you have a remote pic that's stationed in a remote location where he's kind of controlling and observing the actual drone and kind of making sure that all processes and procedures are kind of monitored and making sure that all safety procedures are monitored uh even environmental procedures are modern and I feel like this is one of the good ways of actually working with the uh FAA to get a dvd loss uh cert is to kind of uh write it in a way that semi-autonomous flight is safe it's safer than fully autonomous flight I believe we're not in agriculture we we're not there for full autonomous uh flights but I believe semi-autonomous flights monitored by remote picc in a different location is the bridge and is bridging the gap for full autonomous flights where a drone can create tasks by itself and kind of operate by itself I

**[31:39]** believe we're not there yet but I definitely believe that the semi-autonomous flight is our bridge to getting to this location um I created a chart uh I didn't create the chart I'm sorry I got this chart from the FAA uh and it shows a trend analysis from one of the part 107 31 uh waiver procedure and it shows where it fails and where it succeeds and where people actually need to focus on and answer these questions um I believe they're looking at each uh location or each pv loss application as a unique uh operation for them so agricultural operations in regards to vv loss applying for certs and getting autonomous procedures going on um it is all regards into a the operation that you want these uh procedures to go what you want your autonomous drone to be on and this is in regards to agriculture we have fixed versus rotating locations

**[32:46]** fixed locations are an area where you can geofence a location and that your autonomous operations will always be within this location and will not be moved out of this uh area what this does for the faas kind of show hey this environment is not changing this airspace is not changing um this weather is kind this is this is the area of the region what we need to look at in regards to environmental and obstacles and changes uh rotating locations is where farmers kind of lease land and they move around every uh season uh so they rotate crops in areas so this is kind of a harder uh uh bb loss type of operation to apply for um if you always have a location that's changing um so there's different airspaces you need to attend to and different uh um different uh airspaces different environments different um

**[33:48]** things in weather conditions you have to look at and faas are kind of looking at that as being a hard kind of cell for them but fixed operation is I guess one of the first key I say it will bridge the gap to rotating uh operations but I'm not saying rotating operations are out of the deal because there's certain softwares and certain type of uh checks that can be made uh for bv loss operations that can be done on these other rotating locations so in regards to future technologies that make autonomous drones possible which I believe the agriculture industry is is is needed is is one of the safer um industries that this uh that we will see these type of operations first is basically decision support systems um safety and tasks and regardless of means of safety

**[34:48]** you're talking about utms where we can look at airspace weather and obstacles before we fly so knowing your airspace knowing your weather knowing your obstacles knowing data from adsb to doing what type of uh planes are flying above you is very important software that we need to look at in regards to just decision support systems uh before you actually start an operation and this is for the whole industry in the whole not just agriculture but in regards to agriculture we have task related uh decision support systems so if you saw an irrigation link uh the drone is going to you know create a task to a farmer to create uh to go out to this location capture that data and kind of automate information and we're get where the pipeline is autonomous where now you're triggering different um tasks within the farm uh and

**[35:53]** within the drone so if you fly at 200 feet and an anomaly is um actually seen now you can basically tell the uh tell the drill the drone will know uh based on the data to go lower to go 100 feet to go to 30 feet and kind of capture a lower aerial of this and these will be automated tasks uh that the drone can make or decide on itself based on the data it's collecting so there's levels of data collection that could be automatically fed to the drone that you'll create for itself uh and kind of be the eye on the farm and kind of help the farmer have a hands-on approach without really being there all the time and kind of babysitting the farm where a drone could do this for you and could help manage the farm in a more efficiently and proactive way

**[36:55]** all right so I guess is the question in um answer session thank you guys for all joining here thank you very much darren that was very insightful on how drones are being used in the agriculture industry and we've got a lot of questions lined up from the audience so I'll start one by one so there's a question from jason he's asking are you using gvi or some other non-calibrated rgb vegetation index uh so for the multi-spectral cameras we're using ndvi using ndre we're using multiple indices that the multispectral camera can give off but the software that you process the data with uh is doing these calculations on the data once you're feeding that software these spectral bands of information you can calculate these indices from this information uh I would say uh red edge and dre is one of a special band that definitely can give you a lot of insight to a

**[38:00]** anomaly on the actual field another question on uh there's a question on similar lines do you use ndre to measure chlorophyll content of the country yes ndre is a red edge so we yes we definitely use uh ndre as one of the indices to uh managing a field or looking at a chlorophyll index but comparing that with other indices can probably give you a little bit more insight of what's going on with that actual issue there's another question from dharam he's asking is it possible to measure right the fruit using drones ah so this is the holy grail in regards to data processing uh there is uh research being done right now in regards to different type of crops in regards to citrus in regards to tomato um that are taking images of data and trying to find out the ripeness um I'm not looking at that particularly at this time but I know I do work

**[39:01]** with researchers that are looking and are taking multi-spectral data to find out brightness because it's related it's reliant on when to pick uh a actual farm and some farms uh props are picked in different uh sessions meaning you might have three to four picking um so yes sometimes this is very important and this is research that's being done right now do that in uh another question is is early detection of dc people with recent sensor and drone tank can you repeat that for me one more time sure is early detection of dc is possible with recent sensors and drone technology I think at this time yes uh disease identification uh is possible and you could basically find a disease earlier than someone with a standard eye and also taking it from high resolution from a lower altitude you can basically capture that imaging uh faster than someone scouting so we have done

**[40:05]** research and uh if you look up there's a paper where we actually uh worked with crop consultants where we flew the drone and we had scouts kind of uh scout the field also and we found that at this time you know three to four years ago when this research was done we were 25 more uh precise than the crop consultants in finding anomalies finding diseases because usually people are associated with kind of history they kind of know history on the crop so they usually check in certain areas uh for diseases that if they if they know the actual crop um what the drone was is you know it does give it aerial view so you can see areas that haven't been checked by others um and that can help definitely in uh lead detection in pesticides thank you again darren uh another question is what software is being used to interpret the data to present it to the end client or the end customer so there's uh

**[41:10]** if you look at drone deploy and you look at some of these processing softwares um what there is there's other softwares that are people are building on top of the photogrammetry software to look for these anomalies uh I can't give you right on top of my head there's there's a few out there that are doing stand counts tree counts um I haven't seen anything for autonomous autonomously looking for diseases um but there's there there this is the basis of where the industry is and where I think a lot of focus and software should be is kind of taking uh looking at these type of data to work look at these type of softwares that can create uh triggers or look for diseases or look for some type of analytics on top of the photogrammetry data thank you again there's another question from can you describe premium data and there's a follow-up question how is the imagery

**[42:11]** calibrated and how is time dominant ground truthing performed management decisions can't be made without solid ground truthing I believe you had a slide on on similar lines in the present day okay here can you repeat the first question part of the first part of that question so first is can you describe premium data okay so premium data when I mean premium data I mean the resolution of the data is not distorted uh some issues in and I face this personally is um focus uh issues with the focus on the camera sometimes you can capture and when you look at your screen the actual imagery looks focused but when you look at the data on a computer or inside you have a non-focused image and when you process and give a software engine to process this type of data your map your photogrammetry map is distorted that's one way of looking at the imagery the second thing I'm looking at is environment and weather we're looking at clouds we're looking at um

**[43:17]** any kind of obstacle in the area of the actual photogrammetry area so one for agriculture data you you want you don't want clouds so in regards to going to your next question in regards to calibration so the micasense uh has a panel calibration and has a solar iridescent sensor on the top that kind of take account for uh in-flight um uh calibration to the camera when you are encountered with a cloud but this is uh this is information that the hardware is kind of adjusting in the fly uh when capturing this data and then at the end it kind of uses your calibration panel um and kind of uh creates a medium or kind of flat surface for all your data uh which is is could work and has presented some good type of data for us but I tell you honestly flying without any clouds um and flying between as close as the solar noon 12

**[44:22]** o'clock um is the best for premium data and this is what you want your clients to see is that type of data so there's no issues on the downstream side of ground truthing that data and looking at that data I don't know how much jk I don't know how much I answered of that question yes uh jason I hope that uh I was able to answer your question if you uh more if you need more clarification on that we'll shortly share his contact details so you can reach out to him on those details another question uh is why you why do you prefer to use multi-spectral sensor compared to hyper hyper spectral sensors ah so we have worked with hyperspectral sensors and the only issue I see I see hyperspectral sensors working for research purposes meaning uh for scientific research purposes identifying uh a band or a wavelength that specifically correlates to a disease

**[45:24]** the data is very hard to work with uh you're creating 256 bands of data then you have to process this data you're talking about a heavy workload um I would like I like to work with hyperspectral data when they can give me that brand that works for my multi-spectral so in research they're working right now on finding what specific bands are uh correlated to a specific disease or to a specific anomaly uh the data is very heavy the data is very time consuming the data is more of a research uh level data and it's not really ready for commercial purposes at this time I believe we are about time we can we will take one last question before there's a question from dharam nowadays many commercial satellites claim that they provide centimeter level data with advancing ai and ml algorithm one can do any task

**[46:26]** that can be possible with drone data is there any specific advantage of using drones compared to satellites okay yes so uh in 2017 I've done research with uh the harris satellite so what we did is we had a crop um here about an hour away from us where we put my drone data against a harris satellite and I'm located in the southeast region of the united states in florida george area and if you know this location during the summer we have heavy cloud uh uh skies heavy clouded skies we have rain we have all types of weather kind of conditions it's not like california where you have clear skies I would say 60 70 percent of the time our skies are mostly clouded 60 70 percent of the time and what happened during this research which I guess I maybe have got lucky it was a little bit of a rainy season um there was not one time that a hair satellite was able to penetrate a cloud and give us a

**[47:32]** full uh layout of uh of a field of 150 acre crop um so it was a portion that you know maybe the north side wasn't covered by a cloud or there was a cloud uh in the south side and what happens is when you process that data now you have that big artifact in information which kind of does not help with uh uh tasks um so drones I was able to fly even though sometimes I didn't have artifacts I was able to fly underneath the the cloud and basically you know at 400 feet and create a full map even though there is some artifacts we do have usable data with a drone compared to a satellite that's catching a cloud thank you very much darren uh jason dharam and nick hope we were able to address all your questions if more questions I'm pasting the details where you can reach out to darren and feel free to reach out to him if you have any more questions so darren it

**[48:33]** was a great presentation lots of information uh regarding drones in agri agricultural industry uh before we uh stop any any closing thoughts you you would like to share no I want to thank everybody for coming out and basically uh listening to the presentation and in regards to what I've data I would just like to remind everybody that is collecting eye data make sure you click good collect good data and make sure it goes to the farmer and uh is translated and some type of past is doing it so we can keep on with the industry and we can still be here for a long time thank you very much and uh to all the audience I request there's a networking session that will happen right after this session so request everybody to join in you can also find other presenters and ask them questions one-on-one if if uh if you have any thank you everyone for joining have a good day thank you darren thank you you