Ground Effect Vehicle Long Range FPV Drone
All on its own. This project started back when i built a little foam, foamboard echrono plan and mounted a lidar distance sensor to the bottom to help keep it as low to the ground as possible. The audiopilot flight controller on board is running some custom code that adds pid control loops to the throttle, elevator and flaparons to actively control height based on the lidar reading. The next goal for this project is to add some more custom code that allows it to run autonomous, waypoint missions while in the ground effect a few months ago. I threw this idea out there in another. One of my videos and quite a lot of people seemed excited about it then kevin from the youtube channel thinkflight reached out and offered to build a custom, amphibious ground effects vehicle for sebastian, and i to use for this project kevin just happens to be a master Aircraft designer and builder and hes, also the guy that designed the popular nano goblin, which i can honestly say, is my new favorite fpv wing with skills like that. I was happy to collaborate, so kevin got to work on the airframe and sebastian and i got to work on the flight controller and distance sensors six months later. This is what we ended up with think flight to make it happen. Kevin daniel and sebastian built an rcg eb from scratch. They made the magic happen. Xflr5 epp and hot knife, poor dominated ground effect, helps this airframe fly cabin made it stable cabin made it good its flying on the water just the way it should pink flight, high quality autonomous ground effect, vehicle flying so low its unreal.
The efficiency is down when youre only flying one inch off the ground. Sebastians custom code, written by a pro custom, build ive already pilot got this plane flying low testing endless testing. These boys are never resting, softwares glitching, bugs theyre fixing so much time. They are investing strong little motors, spinning fast, making weight, nose down decent throttle back skim. The leg, bps weight points defined so precisely super fast stabilized efficiency is well that was pretty neat wasnt. It yup so now lets rewind back to the beginning and talk about the journey that it took to get here. This video is sponsored by private internet access. Vpn, pia vpn is the most transparent vpn provider on the market, with a fully open source software and over 30 million downloads they never record or store user data and their no logs policy has been proven multiple times in court. Pia changes your ip address and reroutes. Your internet traffic through an encrypted tunnel, their service hides your online activity from your network administrator online service provider or government sensors. Its super, easy to use – all you have to do is open the app and click connect i use piavpn to access shows on netflix or other streaming services that are otherwise blocked in my country. I love how it safeguards my online privacy by blocking trackers, malicious websites and ads pia is available for all platforms, including ios mac, os windows and linux use my link in the video description to support my channel and get pia vpm for less than three dollars a Month and two extra months for free, that means its actually only two dollars and eight cents a month, thanks to pia vpn for sponsoring this video now back to the project.
If you want to see more on the design and build process, go check out. Kevins video about the project on his youtube channel thinkflight, the majority of the airframe is made out of epp foam covered in kevlar and goop, which is a type of rubbery glue, its mixed with toluene to help thin it out before applying it to the kevlar. The benefits of using goop and toluene instead of normal epoxy resin, is that its much more impact resistant and even a little bit flexible, but it still adds a lot of structural stiffness to the foam. Pretty cool. The wingspan of the aircraft is six feet and it weighs 8.5 pounds. It runs on a 6s 2p 2170 lithium ion pack that powers two little t motor f 90 kv motors with 7 inch props after cavin did some flying of his own. He drove it up from la and i came down from seattle to meet at some random parking lot in portland for the hand off after that. Sebastian and i started experimenting with distant sensors since i wasnt sure, if the little lidar chips that we had previously been using would work over water turns out, they do work, but the range is too small for our new larger aircraft. They are also really hard to waterproof. Since a lot of the clear plastics and tape that i tried actually block infrared light, we tried hooking up a sonar module and got it working with audiopilot.
But those are also really hard to waterproof because its basically just a little speaker and a microphone. They also have a lot of noise in the signal, so we ended up switching back to lidar, but this time a more powerful version. This is the tf mini plus rangefinder module, and it ended up working pretty well for us. I also installed the new dragonlink 900 megahertz system for long range, rc control and live telemetry data transmission on this build were using a cube black flight controller. I spent the first few flights tuning the flight controller and making sure it worked. As expected, i was super impressed with how well this thing flew its really stable and the glide ratio is incredibly efficient for this type of a plane in calm, winds and flat water. It was pretty easy to keep in the ground effect without any lidar assistance, but it definitely requires some precision piloting. One thing kevin explained to me was that, as an aircraft enters the ground effect, the center of pressure shifts backwards. This effectively makes the plane more nose. Heavy, which then makes it want to nose down and run into the ground. This tendency is mostly counteracted by the increase in lift you get while close to the ground, but still the optimum ground effect. Vehicle design would have as little of a center of pressure shift. As possible – and that was one of cavans design goals for this airframe compared to other designs like the air fish, the center of pressure shift is minimal.
Despite the whole center of pressure shift problem, most aircraft are able to stay in the ground effect by themselves. If the throttle is set to the perfect position – and there is no wind or turbulence – this is rarely ever the case, so the next move was for sebastian and i to install the lidar and tune the control loop after a fair amount of trial and error. We got it working pretty well, but the system was still a bit over tuned, as you can see here its interesting how the airspeed and rangefinder altitude kind of oscillate together, but slightly out of phase pretty cool after some more pid tuning. We got it working quite a bit better and i was able to just steer it around like an rc car on cruise control. No elevator or throttle input needed pretty sweet. After that it was time to make it autonomous. We didnt want to risk the new airframe with untested code and luckily the little foamboardy chronoplan was just barely hanging in there. After a lot of abuse, i installed a little gps on it and went out to the park for some testing before testing our new fancy ground effect. Waypoints. I did a normal waypoint mission to make sure the gps was working. Okay, watching this thing fly itself in auto mode was pretty adorable. We also had to verify the lidar worked over grass, so i did a quick flight and manual ground effect mode and it seemed to work pretty well.
Then it was time to test auto ground effect mode. This did not go very well at first. After several crashes we had a smashed nose and a broken wingspar and the thing was just barely able to stay in the air. Oh its got a flappy wing. Now we were just about ready to call it quits, but i decided to do one last flight and miraculously it worked holy its working. This might not look like much, but it successfully flew between two ground effect, waypoints all by itself, shortly before taking a turn too wide and hitting the grass and it crashed it worked, it totally worked. It just ran a little bit too wide on that corner and hit the tall grass thats awesome with the wings about to fold in half the entire time. Thats hilarious. That was good enough for us, so we uploaded the new firmware onto the big bird and set out to the lake again right off the bat. It seemed to work pretty well with this mission. Here. The vehicle would be in the ground effect on a segment, but then climb up to 10 meters for the rest of it. I was too scared of turning while on the ground effect, because we would likely strike a wing tip and crash next. I tried some gradual turns, but it still hit the ground once or twice, which was a bit sketchy, considering that the plane was flying itself. After that i took a break from the autonomous stuff and stuck a dji fpv system on board.
Before installing it, i drenched the whole thing in conformal coding banji seems to like it oh thats, real time and its spooky. Now its in ground effect mode, im just hands off the controller, and it just stays at that altitude. Thats crazy, wait so youre using the sensors to do ground effect right now, so lighter works over chop, your water as well. It seems to yeah thats awesome, all right! Well, thats enough for me very interesting experience: oh theres, a big spider on it. After doing a few more fpv flights, i started having the video cut out randomly and not come back. This was quite terrifying, especially while low to the ground, but luckily i had a return to home switch set up on the radio. After that, i switched over to a full size air unit and that had much better range and didnt randomly cut out. I was able to fly three miles away clear to the other side of the lake, with decent video the whole way pretty impressive. I dont think i would get good video quality from that far away while in the ground effect, but from within a mile away. It seems to work fine, okay, now back to the autonomous stuff, in order to prevent the vehicle from striking a wing tip during turns sebastian made a new type of waypoint that increased the altitude set point to two times whatever it normally was, while in ground effect. If you want oh, oh wow, that hit the water, oh theres, another turn, oh, it gets really low.
On the turns. I tested this out with a big circular waypoint mission and it seemed to work fine at first, but then at one point it started acting funky and hit the ground. Luckily, a guy on a paddle board saw the crash and brought it back to shore the rescue wasnt going super well for him, so he ended up wrapping the leash around it and towing it back. Pretty awesome thanks random guy. The next day i set out to try and measure and compare efficiency in and out of ground effect. I basically designed a waypoint mission where it would fly a rectangle pattern and alternate between ground effect and 10 meters of altitude. This would repeat over and over until i felt like i had enough data after the flight sebastian crunched the numbers in matlab, and this is what we got. The big squares are the averages out of ground effect. It flew faster and drew more current and in ground effect. It flew slower but drew less current, even though this might look like grade a science at first, we actually cant draw any conclusions about efficiency from this data. The problem is that, while the plane is high up, the flight controller is trying to maintain a certain air speed, but, while in ground effect, the airspeed just kind of does what it does based on throttle and pitch, and that is ultimately determined by the lidar input. In order to accurately compare efficiency in and out of ground effect, we would need to keep the airspeed consistent in both modes and compare power consumption.
Alternatively, we could keep the power consistent and look at airspeed anyways. That could be a good test for a future video. One thing we can do is estimate the theoretical efficiency increase based on data from the old aerospace textbooks. This graph shows that, for an airplane of this size, flying 24 inches high will result in a 10 power reduction and flying six inches high will result in a 32 power reduction, pretty cool next up, i planned out some long range autonomous flights and in the process Realized that flying super far away while really low was not a really safe thing to do. It would really suck to lose radio range and then have the thing run into a boat, maybe ill. Try some longer range ground effect missions in the winter when the lake is empty and the water is calmer. But for now i planned the missions so that the aircraft stays in ground effect for the first mile or so and then goes up higher for the rest of the real long range portion. Then, as its coming back, it would go back down into ground effect. I ended up doing this with most of the long range ground effect flights. I did with this aircraft just to be safe. I still wanted to try a long range mission low to the water and luckily ted here was nice enough to take us out on his boat. I programmed in a big mission around lake washington that was nearly 7.
5 miles total. Unfortunately, the morning we chose was pretty windy with really rough water. This ended up being an issue for the rangefinder as youll see in a bit once in auto mode. The plane was not staying as low as it normally would with smoother water. I was manually having to hold a little bit of down elevator to get it to stay lower. Despite this, we were still able to get some pretty sweet, hyperlapse shots. After that we went to the downwind side of the bridge where the water was calmer. Here we did some manual ground effect flights with the insta360 one x2 on board later on ted, took a look at the data and made these two graphs. They show the rangefinder readings in rough water and medium bumpy water. You can clearly see how theres a ton more noise in rough water, not ideal. This is probably why the plane wasnt staying as low as it should have in perfectly calm water. The rangefinder has no noise at all. The line is totally smooth. This all makes total sense, because the lidar beam just reflects in different directions off waves and doesnt make it back to the sensor. Dealing with poor rangefinder data was nothing new early on. We were having a problem where the aircraft wouldnt descend down into ground effect. When the rangefinder was too high to see the ground, luckily these rangefinders actually output a signal. Strength measurement in addition to the distance reading, so sebastian was able to use that to throw out the bad readings anyways enough about the technical stuff lets go to the beach.
So today i brought the ground effect vehicle to the sea and im going to try and do an autonomous mission to a lighthouse thats on that peninsula way over there. This is kind of a risky thing to do because, as you can see, theres boats coming out of that marina right there. So im only going to be in ground effect mode for a little while at the beginning and then its going to go up to 20 meters and do the rest of the mission at 20 meters in altitude. So that should be a bit safer than staying. Super low, where i could potentially hit a boat or a whale. I planned my next mission over the bay in front of seattle so that i could safely get some cool shots of the city without having to fly over anyones head. Look at that just gorgeous. Ironically, the water was smoother out here in the ocean than it was during any of the mornings out on lake washington. This helped a lot while staying in the ground effect. Ultimately, i think the altitude control loop can be improved quite a bit and smoother water also helps a ton so ill, probably revisit this airframe in the winter. Also, if you didnt see my last ground effects video, i would definitely recommend checking it out. We were able to capture some pretty incredible wingtip vortexes on camera, thats all for now, dont forget to subscribe and be sure to check out the thinkflight youtube channel as well.