.png){kind=icon}
.png){kind=icon}
Return
Point cloud-based characterization extraction and trajectory planning for large-scale forgings
.png){kind=icon}
PRE
AI
C
J
T
J
Y
DOI：10.1088/1361-6501/ae5dfa.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This study proposes an intelligent grinding framework for large-scale forgings, with two core innovations addressing key challenges in robotic grinding. First, to solve the trade-off between global optimality and computational efficiency in point cloud registration, a hybrid GoICP-NDT (globally optimal ICP-normal distributions transform) algorithm is introduced. This approach integrates the global search capability of GoICP with the local computational efficiency of NDT, achieving sub-millimeter positioning error (⩽1 mm) while improving registration efficiency by approximately 89% compared to standard GoICP. Second, a regional trajectory planning strategy is developed, which combines multi-scale point cloud segmentation, normal vector-based attitude planning, and tangential stiffness modeling to generate smooth and stable grinding paths on complex surfaces with random defects. Simulation results demonstrate that the proposed system achieves positioning errors within 1 mm in a controlled laboratory environment using a scaled foam workpiece. While these results validate the theoretical feasibility of the approach, we acknowledge that actual grinding conditions—including vibrations, thermal effects, and workpiece material variations—may introduce additional errors not captured in simulation. The ⩽1 mm accuracy should therefore be interpreted as a baseline performance indicator under idealized conditions, requiring further validation in real industrial settings.
Keywords:
intelligent grinding
point cloud registration
trajectory planning
robotic grinding
surface characterization
Journal
IF:
3.4
Papers:
2.6K
Citations:
2.3W
Organization
Cited Papers