The screenshot shows a prototype of a robot simulator. It's a robotarm in a physics simulation which can be controlled manual by the slider in the GUI. The idea is, that the human sets a value and then the signal is transmitted to the simulated servo. The good news is, that the 2D simulation itself is working great. The assumption was, that after programming the prototype, the user will gets fascinated and will test out the GUI many hours similar to a a computer game. But this doesn't happened. Why?

After this short introduction I'd like to give some details. From an abstract point of view, automation is discussed under the term “simulated game”. A real world scenario (which is a robot arm) gets converted into a simulation. In the next step, the simulation is enhanced by Artificial Intelligence. What i don't understand is, why a certain game type makes fun to play and other not. For example a simple snake simulation which can be controlled with arrow keys is a nice game, but if the simulation gets too realistic into the direction of a real life robot arm, the user will loose the motivation to play the simulation longer than 20 seconds.

I've searched the question first in the Internet in advance that other robot simulation games are available. I didn't found them. It seems, that it makes no sense to construct a robot simulator game because the average user prefers a different kind of games. The question is how to formalize this sort of user motivation. Perhaps it has to do with human machine interaction?

enter image description here

Sensemaking by gamification

The robot arm in the screenshot can be controlled by a random generator. This would result into a contentious movement but at the same time it's the opposite of a reward. In games which are played by humans, some kind of goal is available. For example in the Mario AI game, Mario has to rescue the princess and in a soccer game the players have to put the ball into the goal. But in a normal robot simulator such a higher plot is missing.

  • $\begingroup$ This has received a close vote, so it might be worth a revision to make the question more clear. My sense is you are asking about "gamification" that can be used to enhance training of machine learning? $\endgroup$ – DukeZhou Sep 12 at 21:42
  • 1
    $\begingroup$ It is more a cognitive science question than an AI one $\endgroup$ – Basile Starynkevitch Sep 13 at 7:07
  • 1
    $\begingroup$ It looks like there's a kernel of a good question here, but it's hard for me to understand. Are you perhaps asking a question about intrinsic motivation for reinforcement learning? $\endgroup$ – John Doucette Sep 16 at 2:15
  • $\begingroup$ Intrinsically Motivated RL without explicit reward is user oriented. It's about researching why somebody plays a game without having an advantage to do so. The problem with a robot grasping game is, that nobody likes to play the game, because the game itself is boring. On the other hand, the game can't be ignored, because it's a realistic simulation of a real grasping robot. Perhaps it has to do that not all simulations can be gamified. $\endgroup$ – Manuel Rodriguez Sep 16 at 7:54
  • $\begingroup$ @ManuelRodriguez "why somebody plays a game without having an advantage to do so" strongly implies this is a cognitive science question. It does occur to be that the question could be about the qualities of any given game that make the games interesting to humans, but that would be a game design question. Put on hold pending clarifications. $\endgroup$ – DukeZhou Sep 30 at 23:58

Browse other questions tagged or ask your own question.