Complex  robotic systems integrating advanced hardware components and sophisticated algorithms are nowadays  gaining in popularity, thanks to the widespread use of robotic  middlewares such as the Robot Operating System (ROS), whose aim is  to relieve developers from the management of low-level components (i.e., communication, synchronization, drivers). As a direct  consequence, the world of robotics is being approached by domain-experts  with other backgrounds, which are often less interested in the robot as  a system, while focusing more on specific robot capabilities. While they could all greatly contribute to the robotic community, a  considerable amount of effort is still spent in learning how to  efficiently use ROS to develop a fully-fledged robotic system. 
In  this talk, we will present multiple solutions that can be used to simplify and streamline robot  software development. The problem will be approached on two different  levels: high-level applications, such as commanding a robot to patrol an  area, and fine-grained implementation, e.g. embedding existing algorithms in specific ROS components. I will show  the continuity between these two different approaches, not  only outlining how they could benefit from each other, but also how  they could make robotics more accessible for experts from different  fields.