Amazing Open Source Quadruped Capable Of Dynamic Motion

The more we read about [Josh Pieper]’s quadruped, the mjbots quad A0, the more blown away we are by his year of progress on the design. Each part of the robot deserves its own article: from the heavily modified brushless motors (with custom planetary gears) to the custom motor driver designed just for this project.

[Josh], realized early on that the off-the-shelf components like an ODrive just weren’t going to cut it for his application. So he designed his own board, took it through four revisions, and even did thermal and cycle testing on it. He ended up with the compact moteus board. It can pump out 400 Watts of peak power while its 3Mbit control protocol leaves plenty of bandwidth for real time dynamic control.

The motors and gearboxes are also impressive. It took thorough experimenting and taking inspiration from other projects  before he arrived at a 8108 quad copter motor modified and upgraded so heavily its own mother wouldn’t recognize it. This is all packed into a leg unit with three degrees of freedom that puts even the fanciest servo based quadruped to shame.

Finally it’s all packed into a neat four-legged robot frame with batteries and a Pi. You can get a video summary of the robot here or after the break, and we recommend reading his blog for some more images and details.

 

9 thoughts on “Amazing Open Source Quadruped Capable Of Dynamic Motion

  1. The design on those actuators is really impressive! And I agree on the assessment on ODrive, I love the platform but I wish it supported more communication protocols commonly used in industry.

  2. Very interesting design! i wonder if using harmonic drive or cycloidal gearboxes in the shoulder joint will help to reduce energy consumption, because they can’t be drived from the output shaft, like a one way gearbox.

    1. The energy consumption is actually really low for just standing around. It draws around 0.5A in that configuration. With the current 4Ah battery, that’s 8 hours of looking cool. Granted, less when it is moving, but it can still walk for a few hours on a charge.

  3. That sure is a weird walk. Looks like it needs to commit more to moving its centre of balance and gently swinging its legs than just trying to have the legs catch up like a roller system.

    1. Yeah, the gait is just a compromise to get things walking quickly while standing relatively high. I have the legs moving really fast so that the body doesn’t have a chance to tip over much during the flight phase. That’s a limitation of the current, effectively open loop gait control.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.