Grounded language for open system communication

 Robotics in the past was imagined with a computer program as a closed system. The software has to solve a mathematical problem by itself and doesn't communicates with the outside world. The result of this autarky process is the np hard algorithmic problem. Large motion planning problems are overwhelming the processing power of the robot and this results into a robotics failure.

The improved robot control paradigm works with an open system and grounded language as interface to the outside world. The robot is able to execute commands from an operator and reports detected event back the the operator. For doing so the robot is using English which consists of verbs, nouns and adjectives. A typical communication might be "operator: move_forward until junction", "robot: done I've reached junction".

Such an open system isn't affected by the np hard problem, because the robot has no algorithm but the robot executes commands from the outside world. Each command is parsed by the robot in a few milliseconds and the CPU usage is very low. Possible problems like obstacles ahead won't produce a dead end because they are reported back to the human and then the human operator is in charge to bypass the obstacle.

The communication in an open system contradicts existing understanding of Artificial Intelligence. AI in the past was mostly imagined with the previously mentioned closed system paradigm. The robot follows its own algorithms which requires an advanced software and a highly developed computer. It should be mentioned that closed systems in robotics are a dead end. They never worked in the past and its unlikely that future algorithms can address the problems. Closed systems are working only fine for mathematical problems like sorting an array or searching in a database, and they are the wrong paradigm for complex problems like dexterous grasping or biped walking.