Reward Function (Glossary)

The term Reward Function originates from the fields of artificial intelligence, automation, and robotics. A Reward Function, often called a „Belohnungsfunktion“ in German, is a central component in many modern technologies. It helps to control machines or programs by showing them which behaviour is „good“ or „bad“.

You can think of the reward function like a points system. Whenever an artificial intelligence or a robot performs a task, the reward function checks the outcome and awards points – the better the result, the more points it gets. The machine therefore constantly tries to collect as many points as possible in its decisions.

A simple example: a robot vacuum cleaner is meant to clean a flat. The reward function gives it one point for every crumb it picks up. However, if it gets stuck on a chair leg, it gets no points or even negative points. This way, the robot gradually learns how to clean optimally without getting entangled.

In summary: The Reward Function is the reward system with which smart technologies learn to improve and complete tasks more efficiently.