Why can I build working robots but not ones that survive the real world?

You have crossed a real line already. A machine that works is not a small thing, and most people never get that far. What you are hitting now is the gap between clever and robust, and that gap is where the hardest engineering lives. The world does not behave like the bench. Robust design is less about genius and more about respect for failure. The ones who build machines that survive spend most of their time imagining what could go wrong: dust, heat, a loose wire, an operator doing the unexpected. That habit of mind is learnable. You are not missing talent; you are missing the reps of watching things break and asking why. Take your next design and, before you build, list the ten ways it could fail in the real world. Then design against the worst three. Do that every time and robustness stops being a mystery. It becomes a practice you get better at, one honest failure at a time.