Dear Rongzhong Li,
I was one of the backers on Kickstarter for the Robot Dog project and received mine this week. So, I would like congrats you for the quality of the kit and all your efforts for delivering orders. It was certainly a long journey that is not ended for you and your team, but you already have succeeded.
No issue during assembly except for the servo head wire that looks definitively too short whatever the way when screwing to the board, but it was easily replaced by the servo in spare with longer wire. However, I have some questions regarding the software and if you can advise me it would be great before have a look at the code.
I ‘m confronted to few problems. The most recurrent is something that looks to me a bug with the OpenCat code (uploaded from GitHub recently) when gait walk is requested by IR remote. There is no way for stop the robot by press on Balance or whatever button of the remote. This does not happen with other gaits like Crawl or Run that are working perfectly and can be stopped. Moreover, the walk is the sole gait that doesn’t looks smooth, starting with a twitch and the robot is frequently falling despite I have repeated several times the servo calibration. Another point that I observed is a short contact of the front knees with the underside of the frame at each steps.
So could you advise me for fix this problem and improve my user experience. Many thanks.
(picture taken today with my personal robot. Hope you enjoy it.)
I checked step by step libraries and found MPU6050 was different from what I already installed despite it has the same name. Just by changing to the new MPU library the memory occupied moved from 92% to 86% and main issues linked to the overflow on walk gait disappeared. So thank you for your advices
Maybe a dedicated library would help others don't get same problem because erasing all libraries is not convenient when you have lot of other sketches.
About balance posture I observed that Pitch and Roll do not behave the same and would like to know if it’s correct. Pitch axis looks good with extending of front legs while hind are shortening and inverse when balancing at opposite. But for the Roll axis the four legs and shortening simultaneously with the inclination angle whatever the balancing side. I thought that right and left legs should behave at inverse on the Roll axis too.
Let me know if that’s correct.
Thanks
OK. I did it already for the IR remote but maybe it was not taken into account because of other libaries already installed. The current memory occupied after compiling is 92%. So i will check if it reduced.
Thanks
The neck servo is designed with short cables. It will be a little tight when installing but has enough length for the head to rotate.
You need to disable the unused encoders in the IRremote.h, otherwise, they will take too many resources. The "walk" gait happens to use more memory and there will be some overflow and break the normal loop.
The short contact is in the "walk" gait because its amplitude is large. It could be avoided if I reduce the amplitude.