Laurent had his Phoenix walking under PowerPod generated Basic Atom 28 code about midnight France time last night! He is maximizing the code for the new leg geometry, and his intention is to make PowerPod support it fully before he moves on to use the Pro and add more gaits.
I can’t talk about Phoenix without mentioning that I have been under the gun on several important issues dealing with keeping things in stock, so I have not been able to do anything with the parts we had made last year. I am doing my best and hope to have pricing for the unanodized parts soon. I appreciate your patience!
Yes that’s my account. I know Laurent has not been able to finish his work on PowerPod as he is working on Atom Pro IK math anomalies. He is making great progress though.
I feel him on the anomalies. I’m not sure if it’s the same issues since I use floating points in my code, but I found many of it was errors due to ranging problems with the math. I was shoving in ranges beyond what the FACOS, FASIN, FSIN, FCOS can handle…
I tried to use the FATAN/FTAN function as much as I could, since it seems that the tangent functions do not suffer from range limitations like the sine-cosine functions.
Sometimes, the erroneous calculations were so close to what the expected calculations (since it was in radians) that it wasn’t “obvious” at first.
I’ve used about 99999999999999999 lines of “serout s_out” commands for debugging. Matter of fact, my walking video actually spits out my global coordinate (body), local coordinate of each leg, the PWM pulses for each servo since I’ve ran pilot runs while the Phoenix was jacked up on top of a DVD-R case. I then held graphing papers up to each of the leg tips to make sure that the servo was moving the tip in a straight line…
I though it was determined the results from the math were within reasonable accuracies for the platform used. Laurent was using integer math, not FP and found an anomaly. We will post the info soon. Just swamped with the impending move…
Yes, I think the Atom Pro is more then accurate enough for the type of application this is, which is controlling hobby servos. Ultimately, we are limited by the accuracy of the servos, not to mention the calibration, mechanical asymmetry, and etc…
I’m very much a fan of the Atom Pro, as it is a perfect micro controller for my application, plus the IDE is loaded with many functions that are quite handy for robotics and motion control application…
Do you ever plan on moving to Propeller-based Controllers Jim (not that you don’t have 9999999 other plans and have nothing to do )? The Propstick USB has similar form-factor as the atom/pro, bs2, and etc… Would be neat to see a bot-board for the Propstick… Parallax’s protoboard USB is quite nice but it is a bit too bulky compared to the BB…
I am doing my best and hope to have pricing for the unanodized parts soon. I appreciate your patience!