1. Benchmarking of a Camera-less Random Bin Picking Strategy.
- Author
-
Fogt, Tristan, Müller, Alexander, Adler, Timon, Kunz, Holger, and Dietrich, Franz
- Abstract
This paper addresses the limitations of camera-based random bin picking (RBP) strategies by proposing an innovative strategy that relies solely on a 6-axis load cell instead of cameras and image processing algorithms. Specific applications of RBP include loading and unloading industrial assembly machines, transferring objects between industrial workstations, and picking objects to be assembled directly. Camera-based RBP systems are limited in industrial settings due to challenges related to image processing robustness, high costs of hardware and software, the need for 3D models and extensive training datasets. These limitations are inherently tied to the reliance on camera-based technology and image processing AI. This paper presents a novel camera-less RBP strategy, eliminating the need for visual information processing altogether and circumventing these limitations. The strategy is largely agnostic to factors such as bin size, object material and gripper size, and as such can be modularly configured to the user's needs. This strategy is demonstrated using a collaborative robot, highlighting the advantages in a collaborative environment. The proposed strategy is evaluated through feasibility case studies on several objects of different geometries, namely navel oranges, cardboard boxes and resin bottles, and performance metrics are benchmarked. The strategy has an average success rate of 99%, average cycle time of 26.08 seconds per object and an average grip success rate of 47%. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF