ISBN 978-3-8439-5104-3

978-3-8439-5104-3, Reihe Informatik

Tobias Groll
A Hierarchical Approach for Autonomous Planning and Execution of Excavation Tasks

190 Seiten, Dissertation Technische Universität Kaiserslautern (2022), Softcover, A5

Zusammenfassung / Abstract

The present thesis will give a method for solving excavation tasks with autonomous machines. It presents solutions on the different levels of the planning and execution hierarchy of digging jobs. Systems are examined to automatically analyze the given digging job and create a working plan. Furthermore, a control system has been developed for the excavator to execute the given plan autonomously.

A hierarchically distributed solution will be created by dividing a given excavation task in its subproblems. Additionally, mapping techniques are examined to represent the environmental information needed to fulfil the task.

Based on the job description and the environmental analysis, a partitioning of the workspace will be done. Each of the pieces should have a size fitting to the range of the excavator. An approach is examined in the thesis, determining local digging areas that the excavator can handle from a fixed position. By this division, a plan for the excavation work will be determined, including the multiple excavation positions that should be used.

Additionally, within the thesis, an approach is developed to determine the parameters each local digging iteration. It uses rating functions to find the digging attack position according to the present surface configuration.

Furthermore, a autonomous control method is investigated to execute the generated plan autonomously. The approach uses a hierarchical control structure, creating a set of skills that can be used as a toolbox for digging.

A reactive framework is used to implement the skills to enable the systems adapting the execution to unforeseeable events. With this, a continuous adaptation of the execution according to the present environment state is ensured.

Finally, all approaches are brought together into a planning and execution system. This system is tested in experiments using two excavators of different type.