Manipulation task planning and motion control using task relaxations

• Autor(a) principal: Marcos da Silva Pereira
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: http://hdl.handle.net/1843/34528
• topic: Linear temporal logic; Task planning; Constrained motion control; Task relaxations; Engenharia elétrica; Robôs - Movimento; Robôs - Sistemas de controle; Robótica

This master thesis addresses the integration of task planning and motion control in robotic manipulation for tasks that can be relaxed and the generation of feasible manipulation sequences that are executed by a controller that explicitly accounts for the task geometric constraints. To cope with the high dimensionality of the manipulation problem and the complexity of specifying the tasks, we use a multi-layered framework for task and motion planning adapted from the literature. The adapted framework consists of a high-level planner, which generates task plans for linear temporal logic specifications, and a low-level motion controller, based on constrained optimization, that allows defining regions of interest instead of exact locations while being reactive to changes in the workspace. Thus, there is no low-level motion planning time added to the total planning time. Moreover, since there is no replanning phase due to motion planner failures, the robot actions are generated only once for each task because the search for a plan occurs on a static graph. Concerning task relaxation, the task plan action of holding an object toward a target region is relaxed by controlling the end-effector distance to a target plane instead of requiring pose control. This way, instead of requiring six degrees of freedom to control the pose, only one degree of freedom is used to control the distance to a plane. We add constraints that keep the end-effector inside a region of interest and outside a restricted region while it moves toward the plane. Thus, it moves toward the target region that is constrained by the combination of geometric primitives. By using plane constraints, we define a rectangular target region that results in an inverted pyramid trunk region of interest. Additionally, we propose a new interpretation of conic constraints called point-cone constraint that allows defining circular target regions resulting in an inverted cone trunk. We evaluated the adapted framework with pick-and-place tasks with similar complexity to the original framework and showed that the number of plan nodes generated is smaller than the one in the original framework, which implies less total planning time. Lastly, it is shown that the end-effector remains within the regions of interest and moves toward the target region for both the rectangular and circular target regions.