Publication
Navigation Control and Path Planning for Autonomous Mobile Robots
Sebastian Pütz
In: KI - Künstliche Intelligenz, German Journal on Artificial Intelligence - Organ des Fachbereiches "Künstliche Intelligenz" der Gesellschaft für Informatik e.V. (KI), Vol. 37, No. 2-4, Pages 183-186, Springer, 2024.
Abstract
Path planning and motion control of mobile robots are highly dependent on the map representation. Planners and controllers can solve both simple 2D navigation indoors and complex navigation in rough outdoor terrain with multiple levels and varying slopes. Furthermore, an appropriate map representation can enable various approaches. In my dissertation [Pütz S, Navigation control & path planning for autonomous mobile robots. PhD Thesis, Osnabrück University, 2022], I present the robot navigation control system used worldwide in science and industry that can use different map representations while using the same core: Move Base Flex (MBF). It is a mid-level navigation control system that is map representation independent at its core. Building on this, the Mesh Navigation Stack (MeshNav) provides a layered mesh map to represent surfaces in 3D, layer plugins to compute navigability, and global path planners and local motion controllers using the referenced mesh map. The dissertation also provides an overview of path planning approaches and algorithms for various map representations. In addition, the developed Continuous Vector Field Planner (CVP), which computes vector fields for navigation on a 3D mesh map, is presented and evaluated. The developed software packages are open source and are published as binary ROS packages that can be installed via apt. Thus, the provided scientific results can be easily reproduced. Today, the software stack is available in ROS2 and is being further developed and maintained by Nature Robots.