A Framework for Manipulating Deformable Linear Objects by Coherent Point Drift

Manipulation of deformable linear objects is a challenging task for robots. These objects have infinite-dimensional configuration space and are computational-expensive to model, making it difficult for real-time tracking, planning and control. To deal with these challenges, a uniform framework that includes state estimation, task planning, and trajectory planning is proposed in this letter based on the concept of coherent point drift (CPD). A real-time observer is proposed to estimate the states of deformable objects from the perceived point clouds. An online task planner is then developed to recognize the manipulation step according to the state estimation result. For trajectory planning, human operators first train robots example trajectories given several object states. In the test stage, a new feasible trajectory can be autonomously generated by a smooth transformation from training scenarios to test scenarios. A series of rope manipulation experiments on a dual-arm robotic platform are performed to validate the effectiveness of the proposed methods.

[1]  Yunhui Liu,et al.  On the visual deformation servoing of compliant objects: Uncalibrated control methods and experiments , 2014, Int. J. Robotics Res..

[2]  Alan L. Yuille,et al.  Non-Rigid Point Set Registration by Preserving Global and Local Structures , 2016, IEEE Transactions on Image Processing.

[3]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[4]  Andriy Myronenko,et al.  Point Set Registration: Coherent Point Drift , 2009, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[5]  Takashi Suehiro,et al.  In-air Knotting of Rope by a Dual-arm Multi-finger Robot , 2015, 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[6]  Shih-Fu Chang,et al.  Model-Driven Feedforward Prediction for Manipulation of Deformable Objects , 2016, IEEE Transactions on Automation Science and Engineering.

[7]  Dylan Hadfield-Menell,et al.  Unifying scene registration and trajectory optimization for learning from demonstrations with application to manipulation of deformable objects , 2014, 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[8]  Jun Takamatsu,et al.  Knot planning from observation , 2003, 2003 IEEE International Conference on Robotics and Automation (Cat. No.03CH37422).

[9]  Pieter Abbeel,et al.  Learning from Demonstrations Through the Use of Non-rigid Registration , 2013, ISRR.

[10]  Lydia E. Kavraki,et al.  Path planning for deformable linear objects , 2006, IEEE Transactions on Robotics.

[11]  Pieter Abbeel,et al.  Tracking deformable objects with point clouds , 2013, 2013 IEEE International Conference on Robotics and Automation.

[12]  Yunhui Liu,et al.  Automatic 3-D Manipulation of Soft Objects by Robotic Arms With an Adaptive Deformation Model , 2016, IEEE Transactions on Robotics.

[13]  Masayoshi Tomizuka,et al.  State estimation for deformable objects by point registration and dynamic simulation , 2017, 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[14]  Tomaso A. Poggio,et al.  Regularization Theory and Neural Networks Architectures , 1995, Neural Computation.

[15]  Brett Browning,et al.  A survey of robot learning from demonstration , 2009, Robotics Auton. Syst..

[16]  Dan Gazebo Sebagai,et al.  Robot Operating System (ROS) , 2020, Studies in Computational Intelligence.

[17]  Anand Rangarajan,et al.  A new point matching algorithm for non-rigid registration , 2003, Comput. Vis. Image Underst..

[18]  Masayoshi Tomizuka,et al.  Robotic manipulation of deformable objects by tangent space mapping and non-rigid registration , 2016, 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).