Combining task and motion planning for mobile manipulators
- Autores: Aliakbar Akbari
- Directores de la Tesis: Joan Rosell Gratacos (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2019
- Idioma: español
- Tribunal Calificador de la Tesis: Raúl Suárez Feijóo (presid.), Elias Rosello Tamani (secret.), Héctor Geffner (voc.)
- Programa de doctorado: Programa de Doctorado en Automática, Robótica y Visión por la Universidad Politécnica de Catalunya
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
This thesis addresses the combination of task and motion planning which deals with different types of robotic manipulation problems. Manipulation problems are referred to as mobile manipulation, collaborative multiple mobile robots tasks, and even higher dimensional tasks (like bi-manual robots or mobile manipulators). Task and motion planning problems needs to obtain a geometrically feasible manipulation plan through symbolic and geometric search space. The combination of task and motion planning levels has emerged as a challenging issue as the failure leads robots to dead-end tasks due to geometric constraints.
In addition, task planning is combined with physics-based motion planning and information to cope with manipulation tasks in which interactions between robots and objects are required, or also a low-cost feasible plan in terms of power is looked for. Moreover, combining task and motion planning frameworks is enriched by introducing manipulation knowledge. It facilitates the planning process and aids to provide the way of executing symbolic actions.
Combining task and motion planning can be considered under uncertain information and with human-interaction. Uncertainty can be viewed in the initial state of the robot world or the result of symbolic actions. To deal with such issues, contingent-based task and motion planning is proposed using a perception system and human knowledge. Also, robots can ask human for those tasks which are difficult or infeasible for the purpose of collaboration.
An implementation framework to combine different types of task and motion planning is presented. All the required modules and tools are also illustrated. As some task planning algorithms are implemented in Prolog or C++ languages and our geometric reasoner is developed in C++, the flow of information between different languages is explained.
