|PARIGI POLVERINI MATTEO||Cycle: XXX |
Section: Systems and Control
Tutor: BASCETTA LUCA Major Research topic
:Novel Contributions to Robot Force Control for Industrial Manipulators
Advisor: ROCCO PAOLOAbstract:
Control of robot interaction with the environment, generally referred to as robot force control, is required to face the inadequacy of pure motion control for the successful execution of those robot tasks involving contact with a surface. Widely popular since the early 1980s, research on force control algorithms employing a conventional single arm robot has gradually lost its appeal during the last decade, despite the growing employment of robots in finishing and machining operations would strongly benefit from increased controllers’ performance. At the same time, the recent diffusion of new industrial robotic platforms, like dual-arm light weight robots, has driven research on robot force control towards the execution of complex and dexterous robotic tasks, such as bimanual automated assembly.
This project provides contributions in two main areas of robot force control: performance improvement in implicit force control (i.e. an implementation of hybrid force/motion control for position controlled robots) for traditional industrial robots and force controlled bimanual assembly based on trajectory generation for lightweight dual-arm robots.
Force regulation with improved settling performance and absence of force overshoots is achieved by presenting a constrained control approach related to the ideas of invariance control, which is subsequently applied to the implicit robot force control problem. Controller robustness to compliance uncertainties is further addressed. Deterioration of force controller performance connected to environment modeling and identification is prevented employing a data-driven control design approach. An on-line implementation of the controller identification problem is presented, while an outer model predictive controller acting as command governor is introduced to enhance the closed-loop performance.
When a dual-arm lightweight robot is used to perform parts assembly, force controlled bimanual assembly can be treated as a trajectory generation control problem fulfilling force control requirements. A constraint based trajectory generation framework is exploited for this purpose, while estimation of the interaction force enables force sensorless execution of the assembly operation. The presented approach is developed and experimentally validated on a peg-in-hole insertion task and on a cap rotation task.