Comment by scottbez1

Two bits of advice: just jump in, and set reasonable expectations for yourself.

Robotics is a field filled with layers upon layers of complexity, theory, and real-world problems. Those at the top of the game (think robust dog robots, walking bipedal star wars droids, high-speed and high-torque many-DOF arm platforms) are leveraging learnings and expensive prototypes from decades of work/research to achieve those feats. This is not to be discouraging, but to say - they all started somewhere - you too can start, and maybe you won't build things like that today, but you can get there.

And the cool thing is all that robotics work by experts has made the barrier to entry much lower - today you can choose to apply money to solve problems you don't want to learn and focus on the things you do want to learn, in ways that weren't possible 10 years ago. Need a powerful closed loop motor system? Buy an ODrive or closed-loop stepper platform off the shelf, and focus on how to apply those systems to build something higher level.

But the big difference with learning hardware/robotics compared to software (speaking from experience having watched a lot of really smart people struggle to get robots working when I ran an MIT robotics competition, 6.270, and was a lab assistant for another course, 6.141) is that the real world is unforgiving in a way that software is generally not. So if you come from a software background, you will find that robotics can have so many more setbacks because nothing ever moves/behaves/reacts in a precisely predictable way in physical space. This is why I say just jump in - you'll have fun learning all the ways things go wrong, and they faster you get that learning out of the way, the sooner you can build things that work reliably!

Try things and figure out how and why they don't work, and iterate. But don't set unrealistically high expectations or it will just be frustrating.