Home Creators Posts Import Register Favorites Logout

Content

My projects usually start with me saying "that can't be very hard" and eventually end with me saying "wow that was really hard." This project is no exception. 

I've been trying to build a handheld automatic bow that aims and fires for you. My original goal was to aim at stationary targets but my wife told me that was lame. I've upped the ante and have been designing this to hit moving targets. Like my basketball hoops, the crux of this project is having accurate, real time data collection which I can respond to with a fast, precise mechanism.

I've been going through the all the usual motions of integration hell. As is customary I've had a major issue with most parts of the system as well a couple big facepalms of my own creation. But after all that gnashing of teeth I can see the light at the end of the tunnel and things are finally starting to gel. The video I posted above shows a typical shot from the bow with my current software + hardware. 

There are a few things that are keeping me from smashing that target repeatably. The biggest issue right now is timing. To hit the target, my prediction of when and where the target will be has to be accurate to around 40 milliseconds with perfect aiming and timing of the arrow release. If my aiming isn't perfect then that window gets smaller. 

Normally I'd tell you that to a computer 40 milliseconds is an absolute eternity, but the issue I'm dealing with is that the computer is attached to a lot of fidgety hardware that is not well characterized. In the video above the arrow is aimed well, but about 150 milliseconds too late. Things like how long the arrow takes to be released and how fast the arrow flies can easily vary by tens of milliseconds and I'm currently using guesses for pretty much everything. I suspect this is the root of most of my timing issues. 

I'll be spending my Friday night measuring an characterizing a bunch of electromechanical systems. Who needs bars with this kind of nightlife... 

Hopefully it will be working smoothly soon so I can put this video together. I've been excited about this one for several months now and it is so cool to see it coming together! As always I'm very grateful for the support so I can spend time trying to make nonsense like this work.

- Shane

P.S. I got to build yet another catapult...

[UPDATE] Things are now working beautifully. I am so pumped for this video. Here's a sneak peak:

Files

The moving target challenge

Comments

Matt Krohn

Can't wait to see wife mode!

Alex Dowlen

Awesome, Shane! Forget the bars and nightlife; what you do is infinitely more interesting and well worth the effort. Keep up the amazing work!

Xelgium

Sooooo, a handheld, auto-targeting crossbow turret...? 😧

Daniel Whitney

Sounds like a great Friday night! Looks like a cool project glad you went for the complicated build

Ed Reusser

The biggest problem you will have is that a lot of the numbers you are measuring are never going to be certain enough to make it work. You need to figure out a way to use feedback so that you auto-adjust things when the parameters vary. All those measurements depending on muscle strength for example will decline pretty fast.

Ed Reusser

and you can't just tell us you built a new catapult and not SHOW it to us. DUDE! What are you thinking?

John Cox

You have an impressive ability to run fast with a new idea and make _something_ work. I'm on the opposite end: I don't feel like I can move forward until I it's solidly designed. As a consequence, new projects are a real bother to learn. I am gaining inspiration from searching this channel though :)

stuffmadehere

The measured parameters are of excellent quality and I think the non-measured parameters are repeatable enough for it to work. I'm closing the loop on a bunch of things with an accurate whole room absolute measurement system. There's no measurements that depends on highly variable things like muscle strength.

stuffmadehere

Oh don't worry, I didn't hold out. It's what's launching the target in the short video clip at the top.

Jeremy

Have you looked at archery "release" / "release-aid" s? It seems like an automated version would address the "jitter" timing issue, but I suspect it's more complex than that. Maybe some squibs could help with the timing AND up the danger ... slightly.

Colin Fowler

Isn't this basically the "Star Wars" project from the 80's?

Goldie643

As an archer, I am very hyped for this.

Antimaker Space

This looks super cool. Yet again I'm impressed by both the mechanical side but also the math

Rich Gross

This is great, and it’s fun to see the progress toward the goal. Now, the big question for me: How many shots hit the laptop?

Jason

I literally worked on a project for GoogleX/Verily that was basically this. You sir, probably don’t need much stoking of your ego. But I will anyway. Dammmmnnnnnn!!!

Jason

Last I heard, basically, they gave up. (IMHO Too many prefixes and suffixes in the room)

dannydaw59

That's neat but don't freakin kill a beloved living creature like the family cat. It would be interesting what type of targeting system you have on that.

Jason Harner

I'm just gonna throw this out there. Make the target much bigger and it would probably hit it already...

Ryan

Feel bad for anyone who tries to rob you