Beddow, Luke;
(2025)
Learning-based Grasping of Grocery Items using Caging Inspired Gripper Design.
Doctoral thesis (Ph.D), UCL (University College London).
![[thumbnail of Thesis-Luke-Beddow-Robotic-Grasping.pdf]](https://discovery.ucl.ac.uk/style/images/fileicons/text.png) |
Text
Thesis-Luke-Beddow-Robotic-Grasping.pdf - Accepted Version Access restricted to UCL open access staff until 1 February 2026. Download (86MB) |
Abstract
Robotic grasping of varied grocery items, such as for packing online shopping orders, is challenging. This work aimed to develop a grasping approach capable of reliable and stable grocery grasping. Existing gripper designs and learning-based grasping approaches struggle to handle grocery items, which have diverse shapes, packaging, and surface textures. Groceries can be delicate and easily bruised, as well as dynamic and deformable, requiring force feedback. This work presents a novel grasping approach that combines compliant caging with force feedback reinforcement learning. This approach comprises a new compliant caging gripper design equipped with in grasp force sensing, a compliant simulated model of the gripper which has been quantitatively validated, and development of a force feedback reinforcement learning method for training a grasping controller. The novel grasping approach achieved 95.4% grasp success rate on 42 grocery items, with the average grasp able to resist a 3.6N disturbance in the least stable direction, whilst all in grasp contact forces averaged below 2N. Direct simulation to real transfer using the model was shown with an average success rate gap of 3.1%. Grocery grasping performance exceeded all of the most related works, with a median improvement of 11%, and a state of the art 93% bin clearance rate was shown on a grocery grasping benchmark, an improvement of 18%. The design was robust, completing over 5000 grasps over five large scale experiments. Overall, it was found that compliant caging and force feedback reinforcement learning were well suited to reliable and stable grocery grasping. The primary limitation was that performance was tested only for single object and low clutter grasping, and further work would be required to extend this method for grasping in dense clutter.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Learning-based Grasping of Grocery Items using Caging Inspired Gripper Design |
| Language: | English |
| Additional information: | Copyright © The Author 2025. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Computer Science |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10203016 |
Loading...
Loading...Loading...Loading...Archive Staff Only
 |
View Item |
