Why don't they make Huge Fighting Humanoid Robots?

Question

Robotics

All sorts of servo drives, manipulators, precise sensors, cameras, capacious and compact power supplies, compact and powerful computers have been created. But robots capable of acting independently, depending on changes in environmental conditions, have not yet been created. Obviously the problem is in the software.
What are the main problems in programming, after the solution of which it will be possible to manufacture such machines? Image recognition system? Decision-making mechanism in conditions of lack of data? Machine learning?

Answer 1

Because then they will have to be sent to the past.

Answer 2

For starters, humanoid + huge is a very inefficient combination. Very high profile, high center of mass, unstable movement (human walking is a sequence of unfinished falls), if you try to maintain proportions, then there is a problem with the specific pressure on the ground (volume, and hence mass, grow cubed from size, and the area of ​​\u200b\u200bsupport only in a square). Well, if only this huge and humanoid will move in a plastunsky way.
Otherwise, there is only one problem - we need AI capable of independently developing algorithms for performing a task based on its current capabilities and the environment. That is, an assessment of the situation, determining the next step in achieving a given goal, a list of possible actions, predicting the result of these actions, the probability of taking a step and achieving the desired result.

Answer 3

human "meat" is cheaper

Answer 4

I would also add that capacious and compact power sources are only nuclear. And so, even Boston Dynamics use a conventional diesel, because there is still nothing more efficient in terms of energy / mass ratio.

Answer 5

Understanding cause and effect, i.e. the presence of a certain "consciousness" in the machine. But until we know how we think ourselves, the problem looks insoluble.

Answer 6

They've been doing it for a long time, we just don't know about it. No one can be trusted, even squirrels lie.

Answer 7

Let's start from afar, but what do you understand by

able to act independently, depending on changes in environmental conditions

?
To me it looks like:
1) Get and process input. In any format and fast enough to be effective.
2) Transfer the processed data to the perceptron (roughly speaking) and evaluate it.
3) Develop an algorithm based on the results of the previous step.
4) Implement the algorithm, simultaneously updating it for ongoing events.
As far as I can see, there are problems at each of these steps.
I'm sorry, but are you serious? Giant humanoid robots with artificial intelligence?
99% of developers cannot create a primitive mobile application that does not make its users suffer from endless bugs.
I'm not talking about the creation of AI.

Answer 8

YES

Answer 9

Already doing

Answer 10

Why don't they make Huge Fighting Humanoid Robots?

Inefficient. That is, you can, of course, do this, but there are other construction schemes (with a flexible design, for example, in the image of a snake; on caterpillar tracks, etc.)

robots capable of acting independently, depending on changes in environmental conditions, have not yet been created. Obviously the problem is in the software.

The problem, in my opinion, is not only in the software. Why do we need such a robot? What task will he solve? Save people from the rubble? Detain/destroy criminals/violators of the border? Whatever the goal, what happens if the robot is wrong (and it is wrong)? For example, a sentry robot will shoot a lost mushroom picker. Or the rescue robot will not notice a person under the rubble. Who will answer for this? GUI? Software developer? Curator from the border service / Ministry of Emergency Situations? I think it is the issue of responsibility, among others, that determines that at the moment the corresponding robots (for searching for people, for disarming explosives), as far as I know, are essentially "remote hands".