Introduction

Intelligent robots have made significant advances in recent years. Self-driving cars, delivery drones, and security robots are just some of the examples of intelligent robot applications that are becoming viable and where robots have moved into real-world environments.
Furthermore, a lot of robot competitions are used as benchmark problems for robotic research. The complexity of the tasks in those competitions has also increased dramatically. For example, FIRA, the oldest robot soccer competition (founded in 1996), and RoboCup, the largest robot competition for intelligent robots (founded in 1997), started out as a competition of simple differential drive wheeled robots playing soccer using a global vision system. These competitions have evolved dramatically over the last twenty years and now feature 1.4m tall humanoid robots as well as sophisticated wheeled robots using local vision.
The scope of the competition has increased tremendously. FIRA in particular has greatly extended the range of sports for humanoid robots. In 2017, a single robot has to compete in sprint, marathon, obstacle run, spartan race (uneven terrain and wall climbing), long jump, mini-drc, archery, basketball, united soccer, and weight lifting. RoboCup now also includes leagues for home helpers (RoboCup @Home) and work helper (RoboCup @Work, Amazon Picking Challenge) robots.
As the complexity of the environment and the tasks that the robots must perform increases, there is an ever increasing demand for robot middleware to speed up the development of new intelligent robot systems.