Trajectory tracking for the inspection of deformable objects considering manipulability of a 7-DoF serial manipulator

This work deals with the manipulation of deformable objects for defect inspection purposes. The required inspection movements, obtained from mimicking demonstrated trajectories, are realized with a 7-DoF serial arm and a fixed grasp support. These trajectories can be realized in different zones, as such, we use this freedom, with the various possibilities for the inverse kinematics problem and the elbow redundancy, to increase the distance to joint limits and singularities. Thus, we present a novel trajectory tracking method that leads to a high manipulability in each pose of the trajectory, facilitating the implementation of an online correction system. To achieve all this, we use a state-of-the-art approach as a base, but we extend it to consider the continuity of the arm angle in its domain boundary. Moreover, we propose a computer vision-based method to obtain the trajectories. This uses information from the demonstrator's hands to obtain position data and from the deformable object to get orientations. As main results, we could replicate the inspection movements and necessary deformations, achieving a high manipulability in each pose of the followed trajectories.

Automated summary

This work focuses on manipulating deformable objects for defect inspection. By mimicking demonstrated trajectories, the required inspection movements are performed using a 7-DoF serial arm and a fixed grasp support. A novel trajectory tracking method is proposed to achieve high manipulability in each pose of the trajectory, along with a computer vision-based approach for obtaining the trajectories.