Built Dingo Quadruped

Posted by: @kdemotorsports

↑

So it has been over a year since I started this project https://github.com/Yerbert/DingoQuadruped and I finally got it to a stage where it can walk. However shortly thereafter things started to break. I know have some fine tuning to do with the zero position of the legs but I was hoping to find others that have had experience with this particular project. The primary challenge I see with the design is that the legs are attaching directly to a servo and there is only one screw that holds a servo horn onto the servo. There seems to be a lot of stress (understandably) in this area. If anyone else has experience with this design I'd love the feedback on this aspect of the design.

In the meantime you all can enjoy the first 30 seconds of him walking. https://youtube.com/shorts/EHasrskNXck?feature=share  

 

I am impressed you got that far. Your observations re the servos is right on. They are not designed for that kind of load. They are made for planes and drones.

Maybe a gear set between the servo and a final drive that is robust enough for the job.

@kdemotorsports

Posted by: @kdemotorsports

↑

I was hoping to find others that have had experience with this particular project.

Not the Dingo but I spent some time looking at the Bittle robot from Petoi, which appears to  be quite similar.

Posted by: @kdemotorsports

↑

The primary challenge I see with the design is that the legs are attaching directly to a servo and there is only one screw that holds a servo horn onto the servo. There seems to be a lot of stress (understandably) in this area.

Same for the Bittle. I think the bot size and weight plays a significant role here. The Bittle bot is smaller than the Dingo. The Bittle leg design is short and made with a strong plastic. The servos are nylon but have enough torque to flip the bot "head-over-heels" The legs have just 2 joints and have approx. 240-270 degree range. There's a spring "shock absorber" at each joint to help smooth the movement.

The bot runs with a ESP32 on a custom PCB. I spent the most time looking at the software. The postures, gates, and behaviors are lists of servo positions that are either set (with some basic interpolation from the current pose) or run as a loop to get a smoother movement. There's a custom app you get to help create the list of positions or loops with some timing.  These are all saved in non-volatile memory and may be recalled using USB, WiFi, or Bluetooth commands. 

I think because of the size and weight, I didn't notice any real stress of strain except in the head which was designed to come off when stressed. (The result is, of course, nobody uses the head because it's just ornamental.) Otherwise, the legs and joints were pretty stable. FYI, the parts are open source.

In the end, I didn't care for the bot but it was interesting to look at and now I have 2 crash testbots to play with.