Is it important to develop universal humanoid robots?

Why is it important to develop universal humanoid robots?

At first glance, the humanoid form seems expensive and inefficient for robots assigned to perform simple, clear jobs. Asimov addressed this issue exceptionally well. The question is: Is it more efficient to make a specialized, self-driving tractor, or to make a humanoid robot that drives a tractor? Obviously, it is much more efficient and much cheaper to make a self-driving tractor. It doesn't need to be very intelligent, nor does it require so many joints and articulations.

But the question ignores the current ubiquitous reality: we already have millions of tractors that are NOT automated, but are instead built to be operated by humans, ergonomically and functionally. 

All the hundreds of billions of devices around us are intended to be used by humans; including light switches, garage doors, remote controls, shovels, stairs, elevators, bells, pens, phones, traffic lights, buckets, wells, sinks, screw drivers and wrenches.

So, the real question is: Would it be cheaper and more efficient to embed specialized intelligence in each of these billions of objects around us, or would it be cheaper to build a universal robot with a human form, capable of using them ALL?

Yes, it's relevant to build universal humanoid robots.