Clustering Analysis for Secondary Breaking Using a Low-Cost Time-of-Flight Camera
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review
Standard
Clustering Analysis for Secondary Breaking Using a Low-Cost Time-of-Flight Camera. / Niu, Longchuan; Aref, Mohammad M.; Mattila, Jouni.
2018 Ninth International Conference on Intelligent Control and Information Processing (ICICIP). IEEE, 2018. p. 318-324.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review
Harvard
APA
Vancouver
Author
Bibtex - Download
}
RIS (suitable for import to EndNote) - Download
TY - GEN
T1 - Clustering Analysis for Secondary Breaking Using a Low-Cost Time-of-Flight Camera
AU - Niu, Longchuan
AU - Aref, Mohammad M.
AU - Mattila, Jouni
PY - 2018/11
Y1 - 2018/11
N2 - The integration of robust perception in a heavy-duty manipulation control system is an enabler for autonomous mining. This paper aims to analyze performance and robustness of clustering methods for object recognition during the secondary breaking stage of mining. Secondary breaking refers to breaking over-sized rocks into smaller pieces for the purpose of grinding and extraction of valuable ores and minerals. Therefore, recognition of rock pieces is the detection of unstructured targets within a structured environment. The clustering methods are experimentally evaluated by several sets of scenes of point clouds as outputs of a Time-of-Flight camera (ToF). The challenges of rock detection from sparse 3D point cloud data are addressed. In outdoor conditions, ToFs generally provide coarse but robust output in short sample times. Therefore, some clustering methods can be prone to numerical and statistical errors. This paper highlights the weaknesses and strengths of three methods for the secondary breaking application. We propose an algorithmic method for exploiting the existing clustering and segmentation methods efficiently in the detection loop to determine a suitable contact point and approaching angle for a hydraulic jack hammer. The results verify effectiveness of the proposed approach for scattered outputs of low-cost ToFs.
AB - The integration of robust perception in a heavy-duty manipulation control system is an enabler for autonomous mining. This paper aims to analyze performance and robustness of clustering methods for object recognition during the secondary breaking stage of mining. Secondary breaking refers to breaking over-sized rocks into smaller pieces for the purpose of grinding and extraction of valuable ores and minerals. Therefore, recognition of rock pieces is the detection of unstructured targets within a structured environment. The clustering methods are experimentally evaluated by several sets of scenes of point clouds as outputs of a Time-of-Flight camera (ToF). The challenges of rock detection from sparse 3D point cloud data are addressed. In outdoor conditions, ToFs generally provide coarse but robust output in short sample times. Therefore, some clustering methods can be prone to numerical and statistical errors. This paper highlights the weaknesses and strengths of three methods for the secondary breaking application. We propose an algorithmic method for exploiting the existing clustering and segmentation methods efficiently in the detection loop to determine a suitable contact point and approaching angle for a hydraulic jack hammer. The results verify effectiveness of the proposed approach for scattered outputs of low-cost ToFs.
KW - Three-dimensional displays
KW - Rocks
KW - Cameras
KW - Clustering methods
KW - Clustering algorithms
KW - Gaussian mixture model
KW - Range sensing
KW - time-of-flight camera
KW - automatic extraction
KW - 3D point clouds
KW - clustering.
U2 - 10.1109/ICICIP.2018.8606682
DO - 10.1109/ICICIP.2018.8606682
M3 - Conference contribution
SN - 978-1-5386-5861-1
SP - 318
EP - 324
BT - 2018 Ninth International Conference on Intelligent Control and Information Processing (ICICIP)
PB - IEEE
ER -