Hello, I am a lead member of the robotics club at the University of Maryland. We are big fans of ODRI and what the labs have done to democratize this info on legged robots! We are interested in building a quadruped robot, and were interested in building a modification of solo12. However, we have some leftover parts that we would prefer to use instead of buying all new electronics. Also, we would like to have a max payload that gives us freedom to add additional modular components to the robot. We would hope for a 7lb payload, is this possible with solo12? Also, given we are a student club we would like to save money where possible, we have planned a few modifications to save money where possible. The diversions from the original design we were interested in are listed below. (Note: we don’t have as high a standard for precision as a research lab, we really just want it to be able to walk somewhat fluidly, so we thought we could get cheaper alternatives of certain components)
Use ODrive controllers instead of custom microdrive (we have some leftovers)
Use a less expensive IMU
Use a jetson tx2 for the main computer (we have one not in use)
use a less precise/expensive encoder (500-1000cpr)
Standard timing belts/timing hubs off Amazon (we are not sure about this, we read that standard timing belts have near zero backlash with proper tensioning, but we were not sure how significant of a downgrade this would be)
Possible use a higher gear reduction (we would trade slower movement for greater payload, would modify somewhere between 0.8x and 0.5x times the current speed)
Obviously this is a big list and it would be very difficult to predict the performance precisely of the robot with these changes. However, we are more concerned if there are any single changes which are obviously not worth the sacrifice or might not work at all. This is our clubs first venture into legged robotics, and any info/advice/thoughts is greatly appreciated!
Thank you for your time,
Robotics@Maryland
Here are some thoughts – I don’t know for sure because we haven’t tried any of the modifications that you are thinking about.
Switching to ODrive motor controllers and a Jetson computer should work fine if you figure out the communication and if the bandwidth is sufficient.
You will also need to redesign the body structure to make more space for the much larger ODrive boards.
You could use a less expensive IMU- it will most probably require more work to remove noise and drift.
Using a lower resolution encoder will work for motor commutation but it will limit your control performance. That might be fine depending on your requirements and expectations.
You can also use a different timing belt – but that will require some tuning to get the belt tension right. It’s a tradeoff between friction, weight, size and the peak torques that can be transmitted reliably.
The backlash might not be so much of a problem but some of the timing belts have softer teeth and can skip under high loads or impacts.
The Conti Synchroflex AT3 GEN III are the toughest belts that we could find in this size range.
The weight of the Solo12 robot is 5,5lbs so adding a 7lbs payload is a lot!
As you mentioned the gear reduction (especially for the knee joint) should be increased – that can be done by adding a third stage or by increasing the number of teeth on the larger pulley.
All the changes together pretty much require a full redesign of the robot.
I would recommend to design and build one leg and do rigorous testing before building a 12dof quadruped.
I think this is the common problem: it is expensive to build it without proper Funding.
I was trying the same thing, a solo-12 for less than 2,000 usd.
I came to a conclusion that:
Microdriver is cheaper and smaller considering the bandwidth.
IMU -kinda hard to escape expensive ones, unless you use vision or other trick. Noise and drift are not fun.
Lower resolution encoder limit the possibilities too much, you get it working but the trade-off is not being able to explore better controllers and go between simulation to hardware easier.
Payload and the mechanical design can be improved by using Circular Wave Drives. Add Cost and complexity.
I found that I dont have the resources to build the solo-12 so my hope was to find ppl here to trade so I could build 1 leg with those price reductions of bulk orders.
