Intuitive Physical Human-Robot Interaction Using a Parallel Mechanism in a Macro-Mini Architecture
Author | : Nicolas Badeau |
Publisher | : |
Total Pages | : 110 |
Release | : 2020 |
ISBN-10 | : OCLC:1200738514 |
ISBN-13 | : |
Rating | : 4/5 (14 Downloads) |
Book excerpt: This thesis presents the development of a novel macro-mini mechanism allowing intuitive physical human-robot interaction (pHRI). This type of architecture allows the control of a high-impedance robot such as a Cartesian gantry robot in a manufacturing environment using a smaller and lower impedance mechanism, therefore allowing a signi cant reduction of the operator's e ort and fatigue. The proposed macro-mini mechanism consists of a three-axis Cartesian gantry system (i.e. macro mechanism) and a passive three-degree-of-freedom parallel mechanism (i.e. mini mechanism). The mini mechanism is statically balanced at its workspace centre and all three degrees of freedom are decoupled. This means that the gantry axes are individually controlled using the measurement of a single angular encoder of the mini. It also means that the motion of the mini mechanism along the direction of a degree of freedom does not a ect the remaining degrees of freedom, considerably simplifying the control. The use of impedance control with this type of architecture is thoroughly described and analyzed. An experimental comparison with a standard admittance controller using a force sensor is accomplished using a simple peg-in-hole experiment. Results show that the impedance control allows a faster task completion (by a factor of 2) with smaller e ort (by a factor of 20) compared with the admittance controller. A comprehensive stability analysis is also accomplished on several designs of impedance controller, but with the same macro-mini architecture. Results demonstrate that the standard impedance controller is not stable with the proposed architecture and hence an alternative controller is introduced and evaluated. A backdrivable motor is added at the mini's joint in order to render haptic feedback to the operator. Such feedback is used to simulate virtual environment interactions such as walls and collisions with movable objects. The backdrivable motor is also used to vary the impedance felt by the user during control by adding a virtual mass at the mini mechanism end-e ector. Finally, the system's dynamic analysis is used for collision detection of the macro-mini mechanism during planned trajectory motion without the need for force sensors. This last aspect is essential for safe physical human-robot interactions.