Pathplanning for a wheeled robot consists of transferring domain specific knowledge into the software. Requirements like determine the xgoal position and take care of reachability of the robotarm are usually not given as a machine readable description but as expert knowledge of how to manual control the robot with a joystick. What the system engineer has to do is to convert weak defined knowledge into a pathplanner which gets executed autonomously. One option in doing so is a fuzzy control system. In contrast to the famous myth, these systems can be used in complex domains, for example pathplanning.
Quote: “This paper treats the autonomous navigation problem of robotic
systems in a dynamic and uncertain environment. [...] The fuzzy logic
is certainly one of the most adopted approaches in industry.“
Boufera, Fatma, et al. "Fuzzy control system for autonomous navigation
of thymio II mobile robots." Journal of Emerging Technologies in Web
Intelligence 6.1 (2014): 101-105.
Before the pathplanner can be realized with fuzzy-rules and linguistic variables, the requirements have to be written down in natural language. This is part of the software engineering process and is done in expert-interviews. That are humans, how have controlled the robot platform in the past with a joystick and are able to explain which details are important for reaching a position.