| Project/Area Number |
23K03756
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 20020:Robotics and intelligent system-related
|
|---|
| Research Institution | Japan Advanced Institute of Science and Technology |
|---|
Principal Investigator |
CHONG NakYoung 北陸先端科学技術大学院大学, 先端科学技術研究科, 教授 (30362023)
|
|---|
| Project Period (FY) |
2023-04-01 – 2026-03-31
|
| Project Status |
Granted (Fiscal Year 2023)
|
| Budget Amount *help |
¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2025: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2024: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
Fiscal Year 2023: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
|
|---|
| Keywords | manipulation / planar pushing / parameter estimation / multi-contact push / object manipulation / pushing / machine learning |
|---|
| Outline of Research at the Start |
Non-prehensile manipulation aims to find a sequence of actions to maneuver an object given a gripper configuration. This work paves the way toward a new paradigm of contact-support pushing, exploring configuration-varying, multi-point pushes applied on novel objects with unknown inertial properties.
|
| Outline of Annual Research Achievements |
This work relates to the challenges in developing algorithms for robotic non-prehensile manipulation planning. We are exploring ways to find stable pushing directions of polygons of arbitrary shape and size, either convex or concave, using a pair of contact points between a parallel-jaw gripper and two edges of a polygon. We proposed an open-loop pushing algorithm for pure translation that involves only using a vision sensor. We generalized Mason’s voting theorem and applied it to the center of mass estimation of novel objects that suffers from the limited precision and accuracy of the vision sensor and unknown friction of the pusher, polygons, and supporting plane. Our work paves the way toward learning to find stable pushing motions to position and orient objects in a robust and efficient way. We first aimed at generating rectilinear motion, developing the contact normal estimation error bound and pusher motion selection method. It will be followed by curvilinear motion, and eventually a combination of both motions. We submitted our results to top-tier journals and conferences including IEEE T-RO, IEEE T-ASE, and IEEE/RSJ IROS 2024, and published preliminary results in IEEE CASE 2023 and the IFAC Journal of Systems and Control.
|
| Current Status of Research Progress |
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
Reason
We have been carrying out research on object rearrangement (repositioning and/or reorienting novel objects) by planar pushing, which is deemed energy efficient and safer compared with the pick-and-place operation. However, due to the unknown physical properties of the object, re-arranging an object toward the target position is difficult to accomplish. Even though the robotic pusher can benefit from multi-modal sensory data for estimating object dynamics, the exact estimation error bound is still unknown. We therefore examined planar pushing from both theoretical and empirical aspects, focusing on the challenges of using noisy and inaccurate real-world data. We obtained an error bound on the object center of mass (CoM) estimation and extended Mason's voting theorem in the absence of friction and accurate object shape information. The estimated CoM region contains the CoM ground truth in the presence of contact normal estimation error and pushing execution error. We leveraged the estimated CoM region and the zero moment two edge pushing (ZMTEP) method to select the contact configurations capable of tolerating the CoM estimation error for object translation. The object can be translated to the target position with only two pushes at most. Compared with the results of the current state-of-the-art approaches, our results achieved are encouraging.
|
| Strategy for Future Research Activity |
In the following years, curvilinear motions of nearly arbitrary curvatures are aimed to be achieved, endowing a robot manipulator with the capability of learning to perform more complex tasks in the plane by pushing. Primarily, this study serves as a complement to our previously developed zero moment two edge pushing technique designed to achieve rectilinear motions. We initially devise a theory to enable an unknown polygonal object to smoothly track a circular trajectory while remaining in sticking contact with the pusher, irrespective of the friction between the pusher and the object. Similar to our prior work, a variable-stroke parallel-jaw gripper is used as the pusher mounted onto the end of a robot manipulator. We then carry out experiments to validate the theory based on friction estimation. We assess the practicality of the estimated friction forces to find the achievable two-edge contact pushing configurations. Experimental outcomes will provide empirical evidence that confirms the validity of friction estimation for the two-edge contact pushing. Lastly, we find the rotation center to achieve the minimum turning radius, which will help rearrange objects in a confined space. This work lays a solid foundation of planar pushing of objects undergoing curvilinear motions.
|
