The main problem with robotics projects is, that nobody knows how to do it right. So called general Artificial Intelligence wasn't invented yet, and the only technology which is available today are normal computers and programming languages like Python. The pathway from today's computer technology into modern robotics is blocked and the only thing to bypass the obstacle is stage the desired behavior. That means, the engineers can't do science, but they have to do fake science which is called pseudo science or cargo cult science.
It means, to pretend the desired behavior with today's available technology. In the original meaning from Richard Feynman, cargo cult science was some kind of theater play in which airplanes were made out of wood. The reason why is because such wood made airplanes was technology which was available for the people at the island. Transfering this idea into robotics means, that the robot has to controlled with a human operator in the loop. Because telerobotics is the only technology which is known today.
The interesting point is, that lots of technologies are available to realize remote controlled robots. There are servo motors, joysticks, data-gloves and even cameras available. That means, with normal out-of-the-box technology it's possible to setup a teleoperated controlled system. It's not a real robot but has more in common with an airplane made of wood. The airplane is not able to fly and the robot is not able to act autonomously. But it's better than nothing.
Let us describe the details. A teleoperated robot can be controlled with a joystick, a mouse or with a dataglove. It can be controlled with a single person, or with many persons at the same time. A typical lowtech example would be a robot arm which is controlled by two operators at the same who are using a mouse to move the joints. In the optimal case, the human operators are hidden behind paravent because this will make the illusion more realistically. The problem is, that such a robot needs two human operators in the loop. That means, it's productivity is low and the overall setting is fake. But the hope is, such kind of teleoperated robot is inspiring the audience who future technology will look like. It's not a demonstration how well science works, but it shows, that theater playing make sense.
Postpone the hard parts
If a telerobotic system gets activated nothing will happen. The system has no representation of the environment, no planner and no neural networks. The engineers struggled to implement such features because their knowledge is sufficient to solve the task. Instead, a human operator has to send keystrokes to the system and it can't be called a robot, but a trivial mechanical machine.
From an abstract point of view, in a teleoperated robot the most demanding module is missing. It's the AI software which can control the robot by it's own. This module wasn't realized, and the engineers don't care about. They have postpone the step to later moment. They are planning to program the AI in 10 years if their knowledge is greater. Perhaps, they will never figure out how to realize a robot. The robotics engineers have a relaxed understanding of their role. They recognize that the teleoperating robot isn't a real robot but they doing nothing to overcome the issue. Instead, they are done with the project.
Some of the engineers imagined, how to program the AI which is able to control the robot. But they haven't accumulated enough wisdom for a detailed analysis. Instead of realizing the AI part the aim is to deceive the audience and them self and pretend that the robot is working right now. They are doing so by controlling the device manual. From the perspective of efficiency, cargo cult science is a here to stay. Basically the idea is to solve only the easier parts of the project and ignore the harder one. It's equal to a conservative approach in which energy is invested only if it's successful. Instead of trying to fail much often and stand up again, the idea is to never fail and fear about the harder problems. The problem is recognized as too big, and the logical consequence is stay away from it. Fake robotics engineers are staying in the comfort zone and plug the holes with irrational beliefs. A typical statement in a telerobotic project is, that a higher force will control the robot, but in reality the higher force are only two human operators with a joystick in the hand.
No comments:
Post a Comment