Early social robots

 Around the year 2000, the Kismet robot was developed and in 2004 the Leonardo robot was presented to the public. These machines were able to receive and submit emotion and it was unclear why a robot needs such skills. From the self understanding, Kismet and Leonardo were created as social robots with a strong focus on human to machine interaction but their real advantage is located somewhere else.

Classical robotics until 2000 was suffering from the so called state space explosion. Mathematical algorithms like Model predictive control and rapidly-exploring random tree (RRT) were unable to solve motion planning problems. Even simple tasks like programming a line following robot was an advanced multiperson project during this time. The state space explosion problem is the direct result of np hard problems in computer science which is a problem category which needs a high amount of CPU cycles. For example, a robot with 10 DOF has billion of billion possible trajectories to execute and its not possible to score them with an algorithm.

Social robotics is solving the problem by intelligence offloading. Instead of determining the trajectory of the robot by the robot itself, the environment is in charge. A social robot would interact with the human operator, and the operator will provide the needed trajectory. Any complex task which is np hard can be delegated this way, this allows to build minimalist machines which are able to solve complex problems at the same time.