GPS Guided Parachutes For High Altitude Balloons

Most amateur high altitude balloon payloads descend back to earth with a simple non-steerable parachute and can land hundreds of kilometers from the launch site in inaccessible areas. [Yohan Hadji] experienced this first-hand during a balloon launch conducted by his high school, which inspired him to R2Home, a GPS-guided parachute recovery system.

A Teensy runs the show, and controls a pair of sail winch servos pulling the brake lines

[Yohan]’s first challenge was to create a steerable parachute that can deploy reliably, so he started doing tests with a borrowed scale model paragliding wing. He quickly learned that a canopy aspect ratio of below two was needed for reliable deployment, so he started sewing his own canopies. Steering a parachute involves pulling on a pair of brake lines, one for each side of the parachute. A control stroke of about 20 cm was required, and [Yohan] found that RC sailboat winch servos work perfectly for this application. The entire system is designed to fit in a 7×40 cm tube, and the parachute is deployed with the help of a small drogue chute and a servo-operated release mechanism.

[Yohan] is working on a custom flight controller, built around a Teensy 4.1, GPS receiver, and digital compass. A possible alternative is Ardupilot, which we’ve seen used on several autonomous drones, gliders, and rovers. While this system might not be possible to return to the launch point, it could certainly close the gap, and land safely in a designated area.

So far [Yohan] has done a series of test drops from a drone at low altitude to test deployment and steering, using an RC controller. The project is open source, and the mechanical design files and control code is up on GitHub. As with most 16-year-olds, [Yohan]’s resources are limited, so feel free to drop him some financial help on the R2Home GoFundMe page. See the videos after the break for a development montage and project presentation.

The US military has been using a similar system, the Joint Precision Airdrop System (JPADS), since 2006 to deliver supplies to troops in the field. We’ve also seen single-use autonomous gliders developed for the same mission.

32 thoughts on “GPS Guided Parachutes For High Altitude Balloons

  1. Wonder how effective such an idea will be. Wind directions and speeds are far from constant on the way down so it might need some predictive model (perhaps fed actively by the data it gathered on the way up) to know at these altitudes I expect really high wind in x direction, so at the calmer moments before hand deliberately overshoot. I’m sure there are other elements like the topography that could be worth the model knowing too.

    Still got to be better than just drift with the wind no matter what. But I’d love for it to be able to stick the landing on a dinner plate (or even just the same field it was launched from).

    1. This predictive model was done at least a decade ago for a balloon launched glider. For some reason I think the builder lived in Norway, but wherever it was was a place with a lot of places the glider could not land (rivers, lakes, rugged mountains). He made it available for typical balloon launch/recovery to better predict the location for landing based on the amount of time and windspeed at each level that was detected on the way up.

      As I recall the last flight ended on a perfect glide path that was interrupted by a slightly higher mountain on an island only a few dozen meters too tall.

      For the paraglider, the glide ratio is terrible so the best, with modern tech, is to have it land in a place not too difficult to get to. I would doubt it able to ever recover from the down-wind drift on any but a purely calm day. In addition it needs to be able to do so in the face of a headwind. As the glide speed will be low, the ground speed could easily be negative.

      1. Hi !
        >glide speed will be low
        It will mainly depend on wing loading, but reading what you say, I don’t think I’m learning you anything :)
        We can then imagine 2 different strategies, wether the parachute canopy flies fast enough, and the wind is weak enough, and can try to go fly the wind, or wind is too strong, and we have to go down the wind and try to find a landing point on the wind trajectory at ~+- 30° of the wind angle

    2. Hi Foldi-One!

      First Thanks for your interest!

      And Indeed, wind is really something to consider!

      This is how I actually see the problem for a flight on a weather balloon :

      There are online tools that allow you to predict the natural drift of a weather ballon for a given launch location, depending on the weather, (see for example

      So the idea can be to see in general for a launch site what the prediction gives (or even better, retrieve data from real probes over a sufficiently long period of time) and then locate a certain number of “safe zones” that would be on the average drift path of the probes.

      The parachute would then even if there is to much wind, try to not go 100% downwind but maybe 90% downwind an 10% toward a safezone location. On which, if it is known early enough, we could of course place a target or a dinner plate ;)

      About landing on launch fields, it’s something that would be feasible, but on another type of vehicle, (you will guess that it could be on model rocketry), but it’s quite another adventure !

      1. Indeed, I don’t really think a parachute based system could manage return to launch under any but the most ideal conditions. It would just be nice, save all that travel and hunting for it, just sit back with a nice thermos of tea enjoying the outdoors while waiting for it to return.

        The real trick I’ve not seen with any of these recovery type systems is using the data gathered on the way up to verify the pre-launch predictions and select a flight path (and potential alternate landing zone) based on it. But I’m only interested (mostly in model rocketry) not at all a practitioner so maybe there is somebody doing so and I’ve just missed it.

        For myself I’ve doodled and done some back of the envelope precision level maths on something like a Thunderbird 1 system for the final stage (perhaps only stage) of a rocket system – should get a decent glide ratio out of it so return to home. Wondered about the legality of launching such a system from the weather balloon too – so you can fire off that small solid fuel grain and get some fun images of your launch balloon and rocket actually running away from each other against a really cool backdrop.. Utterly pointless musing, having no real desire to build such a thing (yet at least), but it fills the downtime better than crap TV.

        Though seriously the fixed wing (pop out or not) deployment might actually be the best solution for controlled landing so you can reuse the sensor systems – if the whole craft is light enough then a belly landing on the fuselage tube is quite feasible – the impact energy will be pretty low, even if its going to be falling with style at relatively high speed to ground level. Give the wings a touch of dihedral and you should only need one servo acting as a rudder to steer too – so it won’t be that expensive.

  2. Impressive work, especially for a 16 year old! I do wonder if his “slow speed” GPS problem isn’t fixable with a simple predator filter.

    On a tangent I’ve seen HAB recovery methods use fins, gliders, and now a steerable parachute to minimize running about to retrieve the package. Why not a “helicopter” (auto-rotation) technique ? Blades, secured to the tubular body, could be released at the desired altitude and then slow and steer the package towards the desired landing point. It wouldn’t be grid fins and a propulsive landing but still very cool on its own and unique.

    1. Thanks for your interest!!

      >a simple pre[dictor] filter.

      …hmm wait, ‘MacATTACK’ (…) ‘PREDATOR’ help help help !!! anyway xDD

      I should definitely take a closer look at it, I really don’t have any knowledge in the fields of data filter etc..

      >Why not a “helicopter”

      As far as I know helicopter could also be a good idea for recovery, and I know that it has already been experimented with model rocketry, however I’ve always heard that helicopter was not really something that could come under the KISS category :)

  3. Theres some old code from a similar (semi rigid Rogallo wing using atmega168) project here
    This uses logging (to eeprom) during the balloon ascent to measure wind velocity, then uses a time reversed simulation of a (glide ratio of 0) parachute descent from the current altitude to plan an optimal descent trajectory. The simulation process is regularly re-run during descent and used to update the heading control loop. Heading is controlled using just GPS and a single rate gyro, nowadays it’d be worth using some better sensors imo.
    This is 12 years old now, but IIRC it was tested successfully from a steep hill many times, but failed due to a jammed winch servo during flight from 12km altitude. The descent and ascent both saw some quite violent turbulence, something to bear in mind – a parafoil might have been tangled, but the Rogallo wing recovered well, albeit with tangled servo lines as not enough thought was put into the winch design.
    One useful finding: IIRC integrating horizontal movement over 250m height intervals gave the best results for wind measurement, aiui the atmega code at that link uses 100m increments which was found to be somewhat noisier in practice due to GPS errors.

  4. The prediction model for parachute descents for balloons is pretty good now. I would imagine it was pretty easy at descent to know the limits of the landing zone. Then you pick the best location in that area and tell the parachute to head for that.

        1. My hackerspace abandoned the return-to-base glider idea based on our interpretation of FAA rules a decade ago. (UAV out of line of sight.)

          I’m fairly critical of the FAA’s recent power grab below the historical ceiling of 400 ft, but you’re for sure in their jurisdiction if you’re flying around the stratosphere.

          Folks outside the US have different rules.

          We went through with the whole filing-a-NOTAM and so on, doing all of our flights by the book and strictly legal, and it wasn’t that much hassle. (Although planning to avoid restricted airspace in the greater DC area is tricky — we landed dangerously close to Camp David once.) For normal balloon experiments, the system works and is fairly permissive, if bureaucratic.

          But the restrictions on UAVs are odd. It’s not clear to me how a guided glider is more dangerous than an equivalent unguided one. C’est la vie.

    1. Hi NORC,
      This may depend on the interpretation of few things about the category under the parachute come down.
      Whatever happens, weather balloons come down under a parachute, so would that ram-air parachute, which has a relative ability to go in one direction, be considered as a drone, with the FAA I don’t know.
      But in France at least, it (at the moment) seem possible, to, with the authorization to fly a weather balloon, to also fly a relatively controllable parachute on the way back.

      1. Common sense would lean toward it being a parachute. Add a rigid wing structure & it is a glider. But then this is the same agency that freaked out over ultralights a long time ago with a little push from media hysteria. Chances are, they’ve got guidelines hidden somewhere in a 547 page document that was written by 11 lawyers, 23 industry experts and 4 members of Congress, a dead parrot and someone from The Ministry of Silly Walks.

    1. When I researched the idea of any GPS guided device returning to home, I read that in the US, UAV systems beyond 400 feet altitude are against FAA requlations (see I understand this is a UK project and different rules apply. I developed a UAV glider for dropping from high altitude but abandoned it a long time ago for that very reason. If anyone has an update to the ruling please share.

      1. Hi David !
        I’m not based in UK but in France,
        The main difference with your project might be that I’m not using a rigid wing, but a parafoil / raim-air parachute that could be considered not as an UAV (Drone, Rc Airplane, Glider) but as a drag device.
        Also when you launch a weather ballon at least in France for example, you already need a autorisation from the authority, to fly something higher than 150m, so the autorisation to fly this “drag device” could come with the autorisation to launch a weather balloon.
        But that is still something to figure out more in detail, in coordination with the competent authority.

    1. Hi Raul!
      If you are talking about the white square parachute canopy that you can see on the test flight videos, I did it by myself, let me know if you want more information about it.

    1. Hi Joel, Thanks!

      I remember your name from the GPHAR paper author list, when I started looking at prior art a few years ago this was one of the paper that gave me the most helpful informations about what was possible and what was challenging.

      I then decided to experiment with the very low end of payload weights, mostly because the drop test operations would be much easier, but also because it has been less experimented and comes with new challenges.

      But some challenges are the same: I think you have seen we have yet to fly above 3500m, where control gains adjustment in thin air flight is going to be something important. The first very high altitude flight will use a similar approach as used on GPHAR to reduce gain as a non linear function of air pressure. But we will also try navigation at high altitude, which is another challenge.

      If time permits I plan on writing a formal paper on the result of these next tests, with the hope to motivate more people to try things in the field of very high altitude parafoil robots. We’re a very small family at the moment!

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.