Unmanned Aerial Vehicles (Drones) for Measuring Canopy Fuels and Aerial Ignitions

Summary

In this webinar, fire management experts and engineers discuss innovative applications of Unmanned Aerial Vehicles (UAVs), commonly known as drones, in assessing canopy fuels and implementing aerial ignitions for wildfire and prescribed fire management. Patrick, a master's student, shares his research using UAVs to measure tree canopy characteristics and assess wildfire risks. Jim from Drone Amplified presents the Ignis system, an advanced UAV platform for safe and efficient aerial ignition, which reduces risk to personnel and improves fire management efficiency. Discussion includes system capabilities, safety features, regulatory considerations, and potential applications of UAVs in diverse forest environments.

Highlights

• UAVs can estimate canopy cover and tree density accurately, especially in less dense forests. 📏
• The Ignis system from Drone Amplified can safely and efficiently perform aerial ignitions, crucial for managing controlled burns. 🔥
• Field data validation is essential when using UAV-derived measurements for accuracy in forest assessment. ✔️
• Drones offer a unique opportunity to gather data during conditions when traditional methods can't, such as during inversions or at night. 🌙
• There is promising scope for UAV technologies to supplement and enhance traditional fire management operations. 🔄

Key Takeaways

• UAV technology is transforming how we assess wildfire risks by allowing detailed measurement of canopy fuels. 🌳
• The Ignis system provides a safer, more cost-effective method for aerial ignition in fire management, reducing reliance on helicopters. 🚁
• Regulatory frameworks are evolving, providing new opportunities and challenges for integrating UAVs into fire management practices. 📜
• Choosing the right UAV platform depends on specific needs, such as flight stability, cost, and payload types. 🤔
• Future integration of UAV technologies could revolutionize fire management, enhancing safety and efficiency. 🔥

Overview

In the realm of fire science, the integration of Unmanned Aerial Vehicles (UAVs) is proving transformative. During this session hosted by the Southwest Fire Science Consortium, researchers shared insights into how drones are being used especially in assessing wildfire risks through canopy data. Patrick highlighted his work using drones to measure various canopy characteristics like tree density and height to model potential wildfire behavior. His study involves comparing drone-derived data with traditional land management data to improve accuracy and effectiveness.

Meanwhile, Jim Higgins introduced the Ignis system developed by Drone Amplified, an innovative tool for aerial ignition that enhances safety and efficiency. This UAV platform allows for precise ignition in areas otherwise hard to reach, significantly cutting down risks associated with manned aerial operations. The Ignis system integrates seamlessly with commercial drones and offers real-time data to users, making it an invaluable tool for fire management teams striving to manage fuel loads and fire dynamics.

The discussions revealed how the adoption of drones could be both a technological and regulatory frontier in fire management. While UAVs provide breathtaking advances in safety, efficiency, and data gathering, challenges remain in terms of regulation and integration into existing fire management frameworks. Nevertheless, the promise shown by these emerging technologies underscores a potential shift towards drone-assisted fire management practices, heralding a new era of innovation in the field.

Unmanned Aerial Vehicles (Drones) for Measuring Canopy Fuels and Aerial Ignitions Transcription

• 00:00 - 00:30 what you have to say alright great Thank You Xander and just to confirm you can hear me all right over there I can yes thanks for joining we can be loud and clear I can see your presentation so I think you're all set all right great well yeah thanks everybody for giving me this opportunity to present some of this work that I did during my master's here at NAU I think as a grad student it's kind of like your worst fear to do all this work and have it get shoved away in a drawer somewhere and never see the
• 00:30 - 01:00 light of day but this is a really good opportunity for me to ya show some of this stuff so thank you so just kind of a quick preview here on the intro slide I'm so those three images that you see in the middle of the screen are actually all taken in the exact same location in the study area and just kind of represent the different types of ways to look at the same thing here okay so the
• 01:00 - 01:30 motivation behind this research came kind of having to do with all these large wildfires that we see across the west and the land managers response being the fuel reduction treatments to help kind of mitigate these when I was looking at the monitoring plan for a fuel reduction project just outside of Flagstaff and this question stood out to me which was did the investment
• 01:30 - 02:00 effectively reduce the risk of catastrophic fire and so I started thinking how do you even get at answering this question and how do you measure risk of catastrophic fire and so that sent me down the road of fire behavior modeling so in fire behavior modeling it's important to first talk about landfire data which is commonly used to do this sort of thing so land fire is a federal program it primarily uses Landsat satellite imagery and it's
• 02:00 - 02:30 available for free to the public the data is in 30 meter spatial resolution and the temporal resolution or kind of the frequency at which this data becomes available is every two five years and so the product that we use from land fire was the fuels data and so this feels product includes canopy cover which is the percent of canopy within the 30 meter cell canopy height which is the average tree height within that 30 meter cell and then
• 02:30 - 03:00 canopy base height which is the average height of the bottom of the canopy canopy block density which is the density of the canopy topography is elevation slope and aspect and then fuel model which is the type of material that's found within that area being timber or grass etc and so I'm landfire data is what people mostly used to model
• 03:00 - 03:30 fire behavior and so with those fuels products that I just mentioned you take those out of the database and you create what I like to call this raster sandwich and so this raster sandwich goes into the software in this case we used a flame map and the output becomes like what you see on the right there and so what this is saying this is a map and in the blue areas indicate areas that are modeled for surface fire and yellow
• 03:30 - 04:00 areas are passive crown fire and red areas are active crown fire and so this is one way that you can sort of indirectly get at what the fire risk is for an area so general questions that we wanted to answer where can you use UAV surveys to estimate these canopy fuels and then can you take it a step further and use those measurements to model the potential for crown fire and then we
• 04:00 - 04:30 also wanted to see how these different data sources compared when you model the fire behavior now being land fire versus UAV and some more specific questions we wanted to get at where can you use the UAV to estimate canopy cover and then can you delineate individual trees which effectively gives you a measure of tree density and then can you take these individual trees and measure their tree height a canopy base
• 04:30 - 05:00 height and they're both density and so I'll be talking about different data sources and so I just wanted to kind of quickly mention what these are some examples of these so first a satellite imagery and there's all different kinds of satellite satellite data but most commonly used is Landsat soul and scent is 30 meter resolution and the great part about this is that it's free and
• 05:00 - 05:30 next you have aerial imagery or imagery that's taken from an airplane and in this case this is an example from NAIP imagery which is one one meter resolution and is also available for free and so these different examples here are showing the an exact same area so that you can kind of compare them all to each other and third you have UAV imagery and so this is an example of some of our UAV imagery that's in 15 centimeter resolution and so you can see
• 05:30 - 06:00 that there's definitely a more detail when you zoom in there but it definitely it comes out a little bit of a cost compared to the other two free products and then I'll just mention lidar we didn't actually use lidar for this but it's becoming more popular so I just wanted to throw this in here minor is really good at giving you really nice crisp um three-dimensional data but it definitely comes at its own cost so our
• 06:00 - 06:30 study site was located just outside of Flagstaff um that orange rectangle and the pink rectangle are are two different flight areas and combined this made up just over 12 hectares or about 30 30 acres it's primarily ponderosa pine forest um it's it's definitely within the ouya Flagstaff um it's in really short distance to the closest house is right there and I should also mention
• 06:30 - 07:00 that it's part of the Flagstaff watershed protection project which is that local feel reduction project that I mentioned and interestingly this spot right here was actually thin since we did this project so that would be a cool opportunity to go back there and do a similar analysis post-treatment so the UAV or drone that we used is called an EB and it's a fixed-wing UAV it's more of an airplane style rather
• 07:00 - 07:30 than a helicopter style and the sensor on board was a multispectral sensor and so this captures images in four different bands these bands are a little bit different compared to the visible wavelengths or like a photo taken from our like a camera phone or something and this is more geared towards looking at vegetation and vegetation health in particular so our UAV survey did a grid
• 07:30 - 08:00 pattern over that orange polygon and in this grid pattern it had 85 to 90 percent image overlap between the different images that it took the max flight altitude that we set for this was 120 meters which is just below that 500 or 400 foot legal limits for this flight here and the duration was 22 minutes and covered just about 15 acres and these
• 08:00 - 08:30 white points that are on it now represent each photo location and there are nine hundred and sixty images that were acquired in this flight total so with that UAV imagery we created two different products that we used first is the ortho mosaic and so what this is is it's just a combination of those 960 images that are stitched together and basically makes an aerial photo of the
• 08:30 - 09:00 area and so this is that 15 centimeter resolution and includes all four bands that were captured by the sensor and here's an example of what some of that imagery looks like kind of zoomed in a bit so next is a 3d product called a point cloud and in order to create the point cloud we use the process called structure from motion now what this does is it takes different image perspectives
• 09:00 - 09:30 so in that in this example right here if you consider each red point to be a photo location looking down at those trees now you have six different perspectives of those same trees and so the structure promotion software is able to use those different perspectives and create a 3d model of what that would look like and so here's just an example of one of our point clouds from one of those flight areas and you can see here
• 09:30 - 10:00 that the trees were reconstructed in 3d and just for reference this point cloud here has just over five million points so next in our study we also included some field data and this was mostly for validation purposes and we wanted to compare some of our UAV field or UAV measurements to the field data to see how good they were in our field same
• 10:00 - 10:30 thing we did this in a 10 by 10 grid and we picked 57 random plots across this grid and this included seven different density classes because we wanted to include a variety of different density classes from one tree per plot all the way up to seven trees per plot and in total we measured 192 trees and the measurements we took on each tree included its height the base height as well as the canopy diameters so next we
• 10:30 - 11:00 transition into estimating canopy cover to do this we first did this using the UAV data but we also did it using two other remotely sensed data so that we had something to compare to using the UAV data so here's an example of our canopy classification that we did with the UAV imagery so this is in 15 centimeters and the green pixels here
• 11:00 - 11:30 indicate it's canopy whereas black is non canopy then we did the same thing using NAIP imagery now this isn't a 1 meter resolution and then we also use the data set from Zachman and Dixon and this is a data set where they used similar aerial photography to Nate where there's a higher grade photography so as a in a higher resolution and so their product was in a 30 centimeter resolution so
• 11:30 - 12:00 then we took that canopy classification and turned it into a canopy cover percent estimate and so we took um what you see on the right there the green areas of canopy and turned it into 10 by 10 squares and so each 10 by 10 square here represents the number or the amount of green pixels within it and so that's
• 12:00 - 12:30 effectively the canopy cover percentage within that area and then we stack these on top of each other the UAV and nape and the Zachman and Dixon and compared them to each other and we found that the UAV in nape were definitely similar and the UAV and Zachman and Dixon were actually even more similar so next we jump into the individual tree
• 12:30 - 13:00 delineation and so for this you go from the image that you see on the left where it's just a point cloud with your trees in it and you delineate the individual trees out of it so on the right each one of those different colors represent a different tree now this product or this step is needed in order to make individual tree measurements but also to estimate the tree density because then
• 13:00 - 13:30 now you have the number of trees for a specific area so after we did this we did an accuracy assessment and we wanted to see how good we were at delineating the trees versus the GPS locations of those trees that we found in the field and so during this accuracy assessment three things can happen so you have a true positive an omission or a commission so true positive is a one-to-one match so in
• 13:30 - 14:00 this example here the green points represent the GPS few locations of those trees and the pink points are what we found are what we delineate from the UAV point cloud and so we found two trees and there were two trees in the field next you have the omission and so there are two trees in the field but we only found one and then the co mission where there's two trees in the field but we actually found three from the UAV and
• 14:00 - 14:30 through several trials of this the best that we could actually do here we could find 74% of the trees I'm using the UAV data and that's 74% of the trees that we measured in the field and we also had a 16 percent commission so that's 16 percent extra trees so another thing we wanted to take this a little step further and see if the density of the
• 14:30 - 15:00 plot affected our ability to find trees within it and so on the x-axis here on the bottom is the density class or the number of trees within that plot and then the y-axis is the number of trees that we've delineated from the UAV data so I've also included trees per acre and green there which is a little bit more relatable and as far as tree density goes but to kind of walk through this graph here so for one tree for a one
• 15:00 - 15:30 tree plot we're usually able to find one tree from the UAV data from a two tree plot we can usually find two and three for three but the interesting thing here and the main takeaway I think is that when you get into the four five six and seven that they all start to kind of blend together and so what this was telling me is that when you get into kind of those more dense areas and understand in some part of the country some parts of the
• 15:30 - 16:00 country that this might not actually be that dense um but above 160 trees per acre it becomes pretty hard to parse out those trees from the point cloud but below that it seems to do decently so next we wanted to compare the tree heights from the UAV data to those we measured in the field now this is for the trees that we were able to successfully find in the point cloud and when we compared those
• 16:00 - 16:30 heights with the field heights here on the x-axis and the UAV heights on the left or on the y-axis you can see that they're pretty well correlated another interesting thing about tree Heights here is that we learned the Forest Service acceptable error for measuring tree heights in the field and this comes from the FIA protocol and their protocol is less than 20% error and height
• 16:30 - 17:00 measurement 90% of the time and so we wanted to see kind of how we stacked up to that using the UAV derived heights and we found that on average we had a 5.3 percent error ninety percent of the time so for the trees that you can successfully find it does actually measure their height decently well okay so next we jump into canopy base height and to do this we tried using the point cloud height percentiles and so that
• 17:00 - 17:30 image on the right there is a side profile point cloud of a single tree and so the height percentile 5th percentile here means that 5% of the points are below that height 10th percentile means that 10% of the points are below that height and we found that the 5th percentile was the best relationship to or field measured base Heights but it was actually a relatively weak
• 17:30 - 18:00 relationship in the next we tried all sorts of different methods to estimate canopy both density but long story short in the end we couldn't really find any good way to do this that would correlate to what our field measurements estimates would say so why did we try to measure all these I'm kind of just bringing this back and these are all the canopy fuels
• 18:00 - 18:30 estimates that go into modeling crown fire and so in the most basic sense these variables help determine whether you have a surface fire like you see on the left or a crown fire like you see on the right so the last step that we tried was modeling crown fire behavior using the UAV data and now this was kind of more of a trial just to see if that if
• 18:30 - 19:00 it could be possible and to see what the outputs would look like so the different UID data that we plugged into the raster sandwich here was the topography and the elevation slope and aspect canopy cover canopy height canopy base height but canopy block density and fuel model we actually didn't measure using the UAV so we didn't include that so the outputs from
• 19:00 - 19:30 the flame mat fire models um so in this first column here for land fire this is using just landfire data the output said that there is 0% surface fire 14% pasture crown and 86 percent active crown but when we threw everything that we had from the UAV into the model it was drastically different with 100% surface fire canopy base height seemed to be the major player and that were you when we included just canopy base height
• 19:30 - 20:00 from UAV was 98% surface fire so the main takeaway here is and it is possible to use fly map earth used by men with UAV data but it produces very different results and we found the canopy base height was yeah the biggest driver and then implications overall UAV data can be used to estimate some canopy fuels I would say mostly canopy cover it's
• 20:00 - 20:30 pretty good at that it's comparable to the other data sources tree density if your area is not too dense it would actually be pretty decent for this and in the same case for tree heights if you can find those trees then it can measure their heights pretty well canopy base height and canopy bulk density definitely could use some improvement here as far as far as the crown fire modeling goes we found that different
• 20:30 - 21:00 data sources can definitely lead to very different estimates of crown fire and the current data sources in our research we found they can have highly variable accuracies depending on where you're found across the country and you definitely need to consider the accuracies of each of these inputs as you use them it's also important to supplement land fire and not just take it for what it is and typically people
• 21:00 - 21:30 do this with field data as well as expert knowledge for the area but I think UAVs also kind of open up a new a new opportunity to try to do this alright and that's all I got and feel free to reach out to me via email if you have any questions further than what you have here so yeah thank you thank you so much Patrick that's a great presentation really interesting to think how far
• 21:30 - 22:00 we've come questions came in the chat window will press maybe the the three or four that came in and and we'll jump over to our next presentation and there'll be a little bit more time for discussion questions or answers for that at the end of both presentations so I think the first question that came in up was about the sensor on the UAV and Christopher Roberts asked
• 22:00 - 22:30 they're using red edge or tetra kam I'm not too familiar with what the Tetra cam is but this camera included the bands included on that sensor were the red red edge green and near-infrared and do you you don't happen to know the the manufacturer or the the brand of the camp oh yeah I do so that it's called the multi spec 4c okay and so you could
• 22:30 - 23:00 look that up and find different specs on that sensor I will also mention that the equipment that you use both the platt that we use the platform and the sensor this was a few years ago and I think this technology is evolving so fast that there's there's already newer versions of both of those that are more capable than what we use so yeah okay yeah it's it's hard to keep track of all this stuff
• 23:00 - 23:30 similarly Kevin had a question about how you did the image classification and whether you use robos cept or not oh I don't think I'm too familiar with the Robo sect but is he referring to the canopy cover classification yes I think that was that's the disadvantage of doing questions at the end here yes the basically canopy versus non canopy yeah yeah so actually I had the N here's
• 23:30 - 24:00 another slide here so we took the Eartha mosaic that you see here and we used NDVI to classify areas of canopy and actually combine NDVI with the clustering threshold so it had to have a cluster size of a certain number of pixels in order to be classified as a canopy and after that we actually overlaid the digital surface model on
• 24:00 - 24:30 top of that to rule out any areas that said it was canopy but then I really low actually a height of below breast height because we weren't interested in any of the shrubs or the grass we wanted to just find tree canopies above breast height and so including that kind of got rid of a lot of the noise out in the wide open yeah that's really interesting thank you and another question like this
• 24:30 - 25:00 is you sort of just addressed Jeremy's question he asked if canopy cover was based only the off only the ortho or incorporated any structural values like canopy height model and that's like what you were just talking about right yeah yeah exactly and yeah we found that to be pretty effective because um we didn't have a lot of just kind of noise points that were obviously not canopy areas and we found that to be a pretty good way to rule out those okay
• 25:00 - 25:30 we had another question about what benefits of having folks from other parts of the country highlighting our ponderosa pine forests pretty minimal under or mid-story and Kurt is wondering if you can just sort of speculate about how some of your techniques might work in in forest with a more dense a bitter understory I mean they even say gamble oh yeah I think
• 25:30 - 26:00 yeah we were just talking about before would definitely help rule out some of those things and classify between you know your trees as opposed to a shrub but yeah definitely if you include more understory it gets more and more difficult if you have different height classes and all that becomes more and more difficult to parse out and so our
• 26:00 - 26:30 area and I guess relative to the scale of difficulty was probably on then on the end of easier I would say any yeah yeah a couple questions I think we can relatively quickly do you did you look Patrick and how the you have the derive canopy cover canopy high canopy base height compared with
• 26:30 - 27:00 the the landfire that's for the area yeah so I kind of I looked into that just not really as part of this project but more as a personal interest and it's definitely it's hard to compare because you're not it's not really apples to apples they're just resolution differences but they're pretty different even just when you combine our resolution to a 30 meter pixel and average it out we found pretty different
• 27:00 - 27:30 different results okay yeah I mean the general trend was similar like for canopy cover for example if you had areas of high canopy cover they appeared high in both data sets and lo was low but kind of that variation in between was was a little bit different interesting yeah I got sort of points too ten years from now we'll probably have a very different land fibre data set and maybe based on some
• 27:30 - 28:00 of these data collection techniques another let's see your approaches for modeling canopy bulk density kind of talked about in the paper itself oops oh I think I lost you there for a second oh sorry I was wondering us Gregory asked whether your modeling approaches for canopy bulk
• 28:00 - 28:30 density are kind of detailed out in the paper yeah they're detailed out a bit more in the paper kind of in short we use some allometric equations to go from height and diameter to to mass and then we used the canopy diameter to estimate a volume and so having the mass and volume we combined for density and so we did that using our field data and we tried to do the same
• 28:30 - 29:00 process using the UAV data but they just both they didn't match up at all and yeah okay okay great and then the last question was about the NDVI so that the cover map did you just use a simple NDVI threshold or was it sort of a supervised classification to pull out
• 29:00 - 29:30 that part of it so we just use here's another slide kind of I'm having to do with that so we just use the simple NDVI calculation using the two different bands and once we had that NDVI raster that you see here we use the tool called the segmentation image tool and so what this does is you sent an NDVI threshold but also that minimum cluster size that i mentioned and so it has to have there
• 29:30 - 30:00 has to meet that threshold and have neighbors of a certain amount that meet that threshold to be classified out great great well we should change gears to our next Patrick thanks again for presenting and first I should give Jim Higgins the controls so he can start sharing on one second okay
• 30:00 - 30:30 okay so Jim you should have ability to share your presentation and future self etc yeah say that again I said I'm unmuted no yep you are and we can hear you and well Jim is getting his
• 30:30 - 31:00 presentation set up or go ahead and introduce him Jim Higgins is leads the engineering development at drone amplified and he's gonna talk about how UAS aerial ignition device that they've created called Ignace drone amplified if you're not familiar with it is a company that it's a technology company focused
• 31:00 - 31:30 on creative creating innovative tools to provide land and fire managers safe more efficient and more cost effective solutions and really pleased to have Jim on today he previously was research and development engineer at pulse aerospace designing unmanned helicopters and drone amplified and Jim are based in Lincoln Nebraska so it's again kind of cutting across regions here and that's great and
• 31:30 - 32:00 so without further ado Jim I'll turn it over to you and again we'll do some questions and answers at the end thank you okay Zander if you see the the Ignis logo with flames in the background yep it's all working the way should thank you all right good deal all right well I will jump in then as the intermission my slide will advance there we go I'm Jim Higgins with drone amplified and yeah I'm excited to join you guys today thank you very much for having me it's
• 32:00 - 32:30 enjoyable to meet all your new people I see a few familiar names here but a lot of new ones too so I'm just going to talk about our Ignace system we've created today so I'm going to cover some motivation development and capabilities safety the user interface and then hopefully we'll have time for some questions so motivation I'm preaching to the choir with this group but managing fire puts people in danger and it's costly and the US alone
• 32:30 - 33:00 there's 60 billion a year spent and direct costs for managing wild and prescribed fires and an additional 350 billion a year and damages caused sadly the the costs are not just monetary though over the past twenty five years there's been an annual average of 17 firefighters killed while combating these fires and that number appears to be increasing so to help solve this problem we've utilized the growing utility of drones or UAVs and created a
• 33:00 - 33:30 cutting edge state an affordable aerial fire management tool that helps keep personnel out of harm's way so the current solutions for fire management vary in their costs risk and functional capabilities the chart here shows a graph that plots out a few common methods based on the risk and application the drip torch over here on the lower left is useful for small flat areas and carries a relatively low risk the ATV allows more area to be covered
• 33:30 - 34:00 but can lend itself to riskier situations and finally the upper right man helicopter aerial ignitions can ignite the largest and most complex areas but they're also a high risk in terms of personnel and equipment costs our goal at drone amplified was to fill a void in this chart by providing a solution for larger complex areas with a much lower associated risk now we hold no you know delusion that we're gonna replace helicopters because obviously helicopters can ignite much larger areas
• 34:00 - 34:30 but we wanted to fill a gap in this sort of functional chart now talking about the development a little bit of the tool so a brief pictographic history of Ignis is shown here the first several versions were products of graduate research and the Nimbus lab at the University of nebraska-lincoln the system has evolved over about four years from a proof of concept prototype to the commercial product it is today in the lower right
• 34:30 - 35:00 hand corner the video on this next slide will give some details on how Ignis is made this equipment is a technically advanced piece of robotics and automation he was created using software simulation computer-aided design computer-aided manufacturing and computer numerically controlled machine II for the design and development it is made from premium aerospace materials like carbon fiber composites and aircraft grade aluminum more traditional manufacturing is also required [Music]
• 35:00 - 35:30 autonomous systems require significant electronic and sensor integration to monitor and control the various system processes Fitness was engineered to be fully integrated with the UAV platform carrying it for a turnkey solution focusing on aerospace objectives drove part design to be light while maintaining strength and structural rigidity and using a selective laser sintering process for some of these parts results in durable complex geometries at a much lower cost than through traditional machining techniques
• 35:30 - 36:00 the software designed to control Ignace is robust to disturbances and variations and operating scenarios and its focus is on safety all of the technical complexities from the various aspects of mechanical electrical aerospace and software engineering are transparent to the user who interacts with Ignis through its simple and intuitive app for quickly and easily programming ignition emissions [Music] the app provides the critical real-time telemetry necessary for the UAV operator
• 36:00 - 36:30 to monitor and modify missions in progress [Music] the next video here is going to show some of the inner workings of Ignis and how it performs its functions [Music] the operand dropper are secured below the UAV on the payload rails the hopper
• 36:30 - 37:00 is the upper portion that holds the ignition spheres and the dropper is the mechanism below the agitator moves a sphere into position above the loading chute in the hopper and the sphere drops down to the dropper below a hatch opens to allow the sphere to drop into the slipper block in the puncturing chamber once loaded the sphere is forced onto the needle by the puncture motor and drivetrain the hole of the side port needle is now inside the sphere wall so
• 37:00 - 37:30 it's ready for injection the injection motor turns a lead screw depressing the syringe pumping antifreeze through the tube that connects to the back of the needle assembly once the sphere is injected the bottom hatch opens releasing the sphere to drop to the ground below where it ignites 20 to 40 seconds later this process repeats itself alternating back and forth between side a and side B until the mission is complete
• 37:30 - 38:00 [Music] the next slide has a video that just details some higher-level capabilities of the system this video shows the autonomous aerial robotic system for fire management we have created a drone amplified called higness it integrates with commercial UAS or drone platforms it is one of the most technologically advanced new tools for fire management professionals Ignis is easy to use the hopper attaches to the payload rails of the drone and the main
• 38:00 - 38:30 mechanism attaches to the hopper it has a capacity of 150 ignition spheres which are commercially available and already widely accepted one Cygnus is loaded it's ready for flights up to 30 minutes law the system can deliver 30 spheres per minute to quickly ignite burn units once dropped from our innovative system the spheres ignite the fire through a
• 38:30 - 39:00 chemical reaction using our system to remotely distribute the ignition spheres keeps people safe at an affordable price point the custom app we developed includes a mission planning programmed autonomy telemetry and safety features and also captures significant data points for after-action reports there's a tight integration of the mechanical electrical and software engineering and with the help of feedback from fire management professionals we have created the newest piece of safe affordable innovative fire management technology
• 39:00 - 39:30 [Music] let's talk now about some more details of the capabilities hang on just a second all right I apologize for the loud noise I'm in my shop right now and my air compressor just kicked on yeah
• 39:30 - 40:00 sorry capabilities in the system so the the Ignis uses the standard primo fireballs which to the 1/4 inch diameter ignition Spears and then regular antifreeze has pretty quick reload time for a single operator from touchdown to liftoff between missions can be about five to eight minutes it's a lightweight it's about three and a half kilograms fully loaded the wireless communication community RIT communicates over the drones command and
• 40:00 - 40:30 control link so your range should be around up to five kilometers it has manual and programmable triggers and you can actually pre program trajectories and drop points for autonomous missions and then it's delivered in the carry case you see there in the picture so how much can we burn with this system this is one of the most frequent questions we get asked and the answer depends a little bit on the use case but the chart here gives a few typical scenarios that range from burning 50 acres per hour for
• 40:30 - 41:00 a pretty dense coverage with the spheres up to 300 acres if you can drop more sparsely and that depends on your fuel type of course the lower row gives an idea of the total area that can be burned in a day based on the number of hours operated so if you burn for a full 8 hour day you can burn quite a few acres with this move on now to safety safety was a significant consideration when we design Ignace there are over 15 hardware based sensors that measure
• 41:00 - 41:30 everything from motor position motor rate servo position end point limits current draw voltage level temperature response rate communication rates and on and on in addition to these hardware based safety features there are many more software checks running behind the scenes that ensure Ignace is functioning properly and is ready to get rid of a ignition sphere as soon as it's injected it also monitors just the overall operational status of the system as a whole we've also developed a custom app
• 41:30 - 42:00 that provides system specific telemetry as well as a live camera feed and both these serve to increase operator situational awareness the final line of defense in the event of an onboard fire is an emergency release mechanism that can be remotely triggered the picture on the left on the top shows where an aircraft grade steel cable attaches to the drone and then it clips on to the Ignace dropper on the bottom when the
• 42:00 - 42:30 operator triggers the emergency release the dropper disconnects from the hopper and hangs from the tethers had a safe distance while the jammed ignition sphere burns itself out essentially you can see this configuration over here on the right this helps to mitigate the risk of damaging the rest of the system and it prevents the propagation of fire to the rest of the drone I'm going talk about the user interface now which is the app essentially we've created so this is the home screen of the app we developed there are a few different sub menus from here that allow the operator
• 42:30 - 43:00 to download maps and map overlays and also access the the main flight interface so the ability to download those offline maps I just mentioned it's a critical feature that makes our app especially useful for land and fire managers because the internet access is often pretty limited when you're out on site so our app allows operators to access the maps they need while they're at home or the office and they have internet access so they download those
• 43:00 - 43:30 high-resolution satellite images by just pinching and zooming around on the map interface and then clicking downloaded or clicking download rather you can see all the regions you have downloaded on your tablet and these these satellite images automatically populate on the global map within the flight app along with the offline map maps we just talked about the English app supports geo reference pdfs and KML shape files as
• 43:30 - 44:00 overlays on the satellite imagery so for example you can download the days most recent IR map from you know say the NIF CFT FTP site or similar source whatever you use and you can overlay this on your satellite imagery the picture on the right shows what this looks like in the app the UAV position is shown real time during flight over top of that IR map so the operator can correlate the current fire line with the
• 44:00 - 44:30 IR map data from the previous day or they can assure they stay within the boundary of the fireline terrain data is also automatically downloaded for regions saved to the offline maps in the app the picture on the left shows the app screen to play displaying a color graded topographic representation of the landscape you can also see the drone altitude and the home Waypoint altitude and then the terrain data can also be used when you're planning mission
• 44:30 - 45:00 trajectories the app screen on the right shows what points being placed to follow the contours of the landscape this allows for dropping spheres from a consistent height while also maintaining a safe distance above elevation changes in the landscape there's a lot going on in this screen that shows an ignition mission in progress so I'll just walk through at each of the icons and the graphics mean the white polygons here surrounding the area is a geofence that's uploaded to the UAV so that if it
• 45:00 - 45:30 flies outside this area it's gonna stop dropping spheres automatically listen sure the fire is only dropped where you intend to adding a geofence is as easy as just touching three or more points on the map to make a polygon surrounding the burn area the blue and yellow icons are waypoints that are added by touching points on the map they're selectable as either start dropping or stop dropping points the yellow lines indicate drop
• 45:30 - 46:00 lines and the blue lines are transit lines the operator so you just select these lines select whether you want to be dropping lines or transit lines hit the auto takeoff button and start the new view takes off and flies the pre-program it returns Holmwood finished the pink and white spheres trailing the UAV icon show where ignition spheres are dropped and they remain visual visible from
• 46:00 - 46:30 mission to Mission so if you stopped partway through you can pick up right where you leave off on the next mission there are a few camera options [Music]
• 46:30 - 47:00 is my audio back and hear me again okay sorry there are a couple camera options visible and IR the two general categories the visible spectrum on the left and the IR cameras on the right and either of these video feeds can be displayed through the live feed in the app the visible cameras are less expensive and they provide actual extra excellent situational awareness and an
• 47:00 - 47:30 overhead is perspective for analyzing the fire but the one downside to them is as you can see now they get pretty obscured in heavy smoke or at night obviously and that's where the IR the thermal camera really shines the video on the right was shot at night out in Oregon this summer on one of the wildfires and one of the lessons that was learned out there is that Ignis coupled with a thermal camera fills an incredibly useful role because they can
• 47:30 - 48:00 fly at night or during inversions when manned aircraft are grounded and so heating this was used to continue ignition operations at night to help set up for the day crew coming in the next morning and it reportedly made for a very efficient workflow so that is the end of the presentation portion I had today hopefully there are a few questions here how that we have time to answer yeah thank you that does a great
• 48:00 - 48:30 presentation really nice to kind of get into the details cuz I've sort of heard about these things and and now I have much better sense how they actually work so the few first presentation comes in from Pete and he's asking about how you compensate for steady or gusty winds yeah so the UAVs have developed quite a bit over the last you know five to seven years they're quite stable you know here in Nebraska wind is an everyday occurrence
• 48:30 - 49:00 so we we commonly fly in 25 mile an hour winds so certainly you can fly in much windier than you want to do a prescribed fire in and I would say you can still fly on even up to some of the windier days on a on a wildfire they're very very state great and then another question came in about multi mounting multiple cameras could you choose to have both infrared
• 49:00 - 49:30 and visible yes so actually the the thermal video you saw on the presentation is actually a dual it has a visible spectrum camera and a thermal spectrum camera on the same and in our app you just select whether you want to see the thermal image or the visible image or both at the same time so the answer is yes - all okay great you know I can see them both being useful in slightly different situations let's see
• 49:30 - 50:00 Nathan asks about compatibility with other drone sort of bodies the basic drone vehicles UAV vehicles are is the Ignis system sort of custom-made for the the guy or could it be compatible with other UAVs yes it is we have integrated with other commercial systems for better or worse the m600 Pro right now is the
• 50:00 - 50:30 best combination of flight time stability and price the other vehicle we integrated with was about three times the cost flew for a third of the time and was not very stable so you know that value wise the m600 is your best bet right now that being said we can easily pivot and put it on other things I mean we rewrite all the software and stuff ourselves so we can integrate with other stuff if you're if your agency has
• 50:30 - 51:00 something you already use we're more than happy to roll with that and I'll just note I guess it was last month we had another presentation through the same webinar series again on UAVs and there was a whole discussion about DGI and the ability of the federal government to use that platform and I think there's some sort of thing changes there or progress they are in terms of I guess the issue is connection with servers outside the country
• 51:00 - 51:30 so maybe we don't need to get into that now I just want to know that's been come up before and maybe I can share a link to the past webinar as well because folks might be interested in that sure yeah and then a couple more questions here oh I guess I should recognize we are at the top of the hour so I certainly understand if you have to log off we really appreciate all the participants joining today I think we'll probably run over just a couple minutes
• 51:30 - 52:00 to try to answer a few more questions but obviously if you got to go that's fine so Michael asked about a price tag what what what is a system like this cost and it's probably there's probably a whole conversation around but if you can uh sort of ballpark numbers to give people a sense of weirder that's absolutely so the the Ignis payload is gonna be in the fifteen sixteen thousand dollar range because there are sort of a range of configurations you can do but
• 52:00 - 52:30 that will that will give you an igneous payload the m600 which is a DJI product is five thousand dollars you know if you want multiple sets of batteries the battery sets range from 800 to 1200 dollars so I just for an example we sold sort of a complete package to a contract group who basically went from having no
• 52:30 - 53:00 drone anything to having a fully functional package they're burning with on a daily basis and it costs them about twenty five thousand dollars so that that gives you a ballpark estimate you know it'll depend on you what you already have what you need etc but hopefully that gives you an idea yeah I know that's really helpful just to kind of put some numbers on it and then Jeremy asked about delivery of the
• 53:00 - 53:30 thermal maps so available on yes basically the the nifty nifc ftp site is the federal sort of repository I believe for IR map so any incident that they are on where they have daily IR maps you can get those from there I know that a lot of people use like a Vin's a-- and so anything you
• 53:30 - 54:00 can use it of Inza I don't know where I don't know where all people pull their information from but you can pull it from there okay and then Paul was asking about safe return to launch point if there are problems or contact is lost or you know something doesn't cause plan is that search here for the system yes yeah so it's a it's a pre-programmed basically response so before you take
• 54:00 - 54:30 off you would set your behavior if your battery runs too low or if you lose connection you would say you know come home at 100 meters height or you know whatever safe height is for your location you would pre-program that and it would it would come home when it's done okay and then Craig asks about the flight control are there there options to use other flight control software or
• 54:30 - 55:00 does it need to be the the Ignis software because the the dropper integration right there there's no other software out there right now that will run you know the ignition processes there are many apps for flying UAVs but if you want to do ignitions with higness you and you need to use our app yes yeah okay and just to clarify Jeremy was
• 55:00 - 55:30 asking about the geo referencing and so all the the flight paths and the drops are all geo-referenced right yes that's correct okay so sort of following up or looking a little out of order I should have found this question sooner but back to the issue of wind so is it possible to do any adjustments for a wind effect on the spheres as they drop
• 55:30 - 56:00 I guess the idea being if you're if you're going and some heavy winds like you guys happen to braska do you plan drops differently so that the balls you know move as they drop yeah that would sort of be a pre-planning thing so as you're programming your mission you know if you know you have a very strong east wind you don't want to fly a little bit to the west of where you know you wanted them to drop it is not doing any in-flight calculation wind is a very challenging thing to measure especially
• 56:00 - 56:30 from a rotor wing vehicle that's creating significant wind itself so we haven't tried to tackle that technical challenge yet so yeah so you just sort of need to adjust your flight path based on what you feel the conditions are okay let's see we actually we have a bunch of questions here I'm gonna try to pick out
• 56:30 - 57:00 some ones we haven't necessarily talked about I guess one I think this is coming from a Canadian colleague are there you want to talk at all Jim about the sort of restrict restrictions or interactions in terms of regulatory constraints on doing aerial ignitions yes so I see that Brad carrots just weighed in a little bit there are significant regulations
• 57:00 - 57:30 here some agencies have their own agreements so in the States at least I should talk I'm I'm not as familiar with Canadian policy but in the States certain agencies have agreements with the FAA that allow them to do certain things for us as a drone amplified as a private company we had to go through a couple different processes basically getting waivers for certain regulations
• 57:30 - 58:00 that the FAA put forth to allow us to drop incendiary devices and I you know I would lat I would let you know Brad or Steve some of the guys from the OAS group weigh in on what all agreements they have but they they are allowed to do some things that we are not allowed to do just because as part of the federal federal government they're tied in a little closer and have you know pre-arranged agreements in Canada I I don't want to weigh in
• 58:00 - 58:30 because I'm not nearly as familiar but I would certainly imagine there are some you know analogous regulations that you would have to work through yeah yeah I know that all makes sense and I you know obviously certainly the consortium is not advocating people like you know gonna draw up you know there there are paths all on these things but I think our motivation here for for inviting the presentation was really like this this
• 58:30 - 59:00 is a change in technology and I know thinking about trying to get good fire on the ground in the southwest there are some barriers that this could overcome them and that's the sort of context I think it's really important to think about hey are there there there burn units or opportunities that were missing that we could if we follow the right approval process etc could we address them with these sorts of systems but I think it's it's important to highlight
• 59:00 - 59:30 that yes there are rules here yes absolutely let's see yeah and someone actually suggest a policy webinar and that's that Thank You Bradley that that could be really useful clearly there's an interest in in UAVs and maybe I'll follow up with you Bradley and sort of think about how the best way to do that may be well I'll pick out one or two more questions here and then we'll sort
• 59:30 - 60:00 of wrap up David was asking you you sort of talked a little bit about purchase price Jim guys have any kind of estimates obviously depends on how many balls are dropping per minute or acre but sort of an hourly cost of operation and I guess I often hear helicopter hourly costs and so maybe that's the kind of comparison we're hoping for right you know there is
• 60:00 - 60:30 an electric power vehicle so you know your electricity rates vary upon region if you're running a generator you know I your your consumable cost is pretty low antifreeze is cheap the spheres are a little bit more expensive but you know I would say there's probably an annual you'd have to break down the annual maintenance cost into the hours per year you use so I guess in
• 60:30 - 61:00 short I don't have a great answer for you because it doesn't depend a little bit on you you know the batteries you use the batteries you should SPECT if you take good care of the batteries you should get at least you know at least 400 charge cycles out of those I'd say closer to five or six hundred so yeah I mean it's fairly low I would say comparatively to many other tools you'll find yeah and I would hazard that things
• 61:00 - 61:30 like staffing cause like once you what if you capitalize like capitalize a purchase cost and a lot of your operational hourly costs are gonna be things like staffing yeah definitely good conversation an important question Chad asked about have you guys considered modifying the firing mechanism for the 26 millimeter dragon egg spheres yes
• 61:30 - 62:00 so we originally went with the the fireballs the the thirty two millimeter ones because most of the people we talked to already used those for their helicopter ignitions so they said oh yeah we got ten boxes of those in the shop right now so that lower the entry bar for us a little bit by using what most people were already familiar with we are actually currently working on a design for the smaller dragon eggs to basically increase capacity great and then maybe our last
• 62:00 - 62:30 question here Christine is asking about smoke monitors and you know whether you you guys have thought about her or whether it might be possible to attach a smoke monitor particulate matter meter so you could measure that during ignitions as well I would say probably I'm not super familiar with smoke monitor technology
• 62:30 - 63:00 but you know if it if it's a fairly small light sensor it seems like that would be decently easy to integrate the only thing I could think of is that you know rotor-wing aircraft create a fairly significant downward breeze so if you need real still air you might have a little bit of a hard time finding that in the very close vicinity to the props but if that's something that's really useful we I'm certain we could figure something out ya know that's it's an
• 63:00 - 63:30 interesting idea I know we've been thinking a little bit about smoke monitoring because that's often a real hassle with prescribed fire making sure that we're not impacting people's health so okay actually I just I got a text from one of the OAS guys who said they did it with volcano particulate matter in Hawaii so no yeah cool cool all right well there we go all kinds of new topics here cropping up well I want to respect
• 63:30 - 64:00 everybody's time Jim I really want to thank you Patrick thanks again to you for your presentation I know it was a little bit there's a lot of information today and which is why we went over but I think it's fun to kind of combine these two uses for UAVs in one presentation so thanks everybody for sticking around and listening and I hope you'll join us on future webinars so thanks everybody very much thank you