That's impressive... just out of interest, am I correct in guessing the raspi would be too slow because of the image processing required? Do you have any feel of how long would it take to process one frame? I suppose the question is, what spec of PC are you currently experimenting with to get 30fps...tom.larkworthy wrote:Just stumbled across this thread. Hi! I am in the process of adding optical feedback for this device. I am 80% confident it can reach 0.1mm accurate. I am using libusb and programming it through python, and using pulse width modulation to slow the arm down. Unfortunately the RaspberryPI is not fast enough to do the image analysis at 30fps, so everything except the controller needs to be done on a desktop (I just use desktop)
please see http://edinburghhacklab.com/2012/05/opt ... -problemo/ for more details.
Presently I am integrating adaptive feedback control (http://wcms.inf.ed.ac.uk/ipab/slmc/rese ... tware-lwpr) and hope to have another blog post out in a couple of months. (the control is tricky)
Yeah I saw "The Thing" when you posted it, I liked the way you measured the main rotation (waist) using a roller on a multiturn pot. That's currently on my (long) list of things to try someday...jekw wrote: A blog post, http://jekw.co.uk/2014/01/05/the-thing/, gives more details and shows the grip/clamp I'm using to pick up a (mouse) ball instead of trying to control the original gripper accurately.
Users browsing this forum: No registered users and 7 guests