RG-Trees: Nonlinear and deterministic extensions to reference governor trees

Date

2025-2-27

Author

Gölbol, Ferhat

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

109
views

0
downloads

Cite This

Obstacle avoidance motion planning is a basic problem in mobile robotics. In some cases, e.g. in a factory, map of the workspace is known a-priori. The planner's main task is to find an appropriate feedback control law that takes the robot to the goal location without colliding the obstacles. Previously, Golbol et al. proposed RG-Trees, an OAMP algorithm that combines random sequential composition and the Reference Governors (RG). They covered the obstacle-free space with overlapping square-shaped safe regions, and steered the robot from one region to the next one, staying inside the safe region, using RG. A major drawback of this algorithm is that it is only applicable to LTI systems. In this work, we propose replacing the RG part by the Explicit Reference Governor (ERG), extending its range of application to nonlinear systems. Moreover, we propose three deterministic tree generation methods based on Voronoi diagrams to replace the random algorithm. In this way, we obtain the same, `optimal' path in each run. We develop three nonlinear motion controllers, for Unmanned Surface Vehicle, differential drive robot and dynamic car models, and verified our algorithms on them.

Subject Keywords

Motion planning, Obstacle avoidance, Reference governor, Voronoi, Deterministic

URI

https://hdl.handle.net/11511/114220

Collections

Graduate School of Natural and Applied Sciences, Thesis