My IMU stabilized bike has yet to (intentionally) leave the ground, so it doesn't get very far away from me. I'm not sure if this would be useful to those of you with drones - definitely not planes, but maybe quad rotors could benefit?
From the beginning, I've been sending data from the bike to my laptop and logging that data so I have some idea of what's going on. It's been useful for bench testing as I can lean the bike and see the graphics on the screen change or I can watch the noise on the accels when the motor runs and see that I need to adjust the filtering, etc...
The problem is when I go to run the bike and see something funny happen, I can't go back through the data and pinpoint the exact instance that the glitch occurred. I don't know when the bike is in a turn if it really knows the current lean angle or is the lateral acceleration throwing off the DCM at the time. I can try and watch the laptop screen and the bike at the same time, but that hasn't worked very well. (Though this does seem like a good use for those Epson Moverio glasses. hmm..)
Well, this is the solution I've come up with for now. The bike sends data to the laptop using a pair of Xbee transceivers (I can also send data back to the bike - currently just using this to update the PID constants). The laptop logs this data to a file but also sends it to another application that is capturing video from a Dazzle USB frame grabber. Using a DirectShow overlay filter the video capture application adds a time stamp and a limited subset of the data to the video frame. Using this I can now go back and review the video footage along with all of the data that was captured during that frame.
Now I just need to wait for the weather to improve so I can get back outside. I also need to look into a head mounted camera so I don't have to bug a friend to help out.
No doubt this has all already been done before, just thought I'd share...
Dave
Comments
I did move the front wheel behind the forks to increase the trail which seemed to help when running off road. On road it wasn't a problem with less trail. If you ever seen a "normal" r/c bike the trail is much greater - the triples are actually reversed with the pivot point leading the fork tubes. Almost like a caster wheel, but they depend on the geometry self balancing the bike and the steering servo just pushes on weak springs to apply a small torque to force a lean.
As far as counter steering goes, currently the control is always counter steering. I know there is a speed under which counter steering shouldn't be necessary, but for now it doesn't switch to non-counter steering at slower speeds. Speed is definitely taken into account when determining how far to forks however. I've wondered before if a little scale bike like this would be helpful in bike/motorcycle research - seems a lot cheaper and safer to experiment on a model vs the real thing.
Kudos, just what I have bee searching for.