frail cosmos
Hello
I have been working on a quadruped robot the mechanical is sound and I also have a policy that I am reasonable happy with.
My last hurdle is now taking this policy loading it on the jetson and running it how do I do this if someone can point me towards some good recourses that go through this ect I though this would be the easy part but it seems quite all over the place.
Thankyou in advance
tawdry light
@frail cosmos this is beyond where I am, but I did find this Sim-to-Real Policy Transfer guide.
This Transferring Robot Learning Policies From Simulation to Reality guide in the Robotics Fundamentals learning path covers details about why this is the hardest part.
Hopefully that helps you get there!
frail cosmos
Yeah it's a good start thankyou I do find that all these thing lack visual examination like what the software stack looks like the code in each file what it does what to download in each order ect
tawdry light
I get that, @frail cosmos. My guess is that's due to a wide range of variation, which is tricky to document. Although, I did some more digging and found Deploying the trained RL policy on Spot with Jetson Orin, which might be exactly what you're looking for.
Tag me to keep me updated on your progress!
steep dawn
for sim2real id recommend reading papaers as well on locomotion policies
there should be a lot on quadrupeds and you can copy their deployment methodology
same with sysid and DR to make sure sim2real is proper
frail cosmos
I get that, <@664255830442508298>. My guess is that's due to a wide range of var...
will do