Icon

2023-Valtorta-Riconoscimento anomalie picknplace cobot

There has been no title set for this workflow's metadata.

This workflow shows an implementation of anomaly detection for a robotic arm based on data collected from an external sensor.
The source dataset, as well as the original Jupyter implementation can be found at the indicated link.

URL: github.com/hkayann/CASPER-PerCom/blob/main/casper.ipynb https://github.com/hkayann/CASPER-PerCom/blob/main/casper.ipynb

Anomaly detection Load and clean data from nicla sensor. Loading data from right arm. Reading the nicla.csv external sensorfile first, skipping the first 6 linesdue to faulty format/data.Visualizing data,we see the sensor idlingfor some time.By selecting column with |GyroX|>1we find that the sensor start workingfrom row 898, thus we ingore the first 897rows.We consider rows starting from row 898 to row 1728006to match right arm data.Node 10Resetting timestamps.Getting first row.Converting first timestamp to var.Selecting onlyanomlay states.Finding last occurenceof anomaly state.Cleaning last faultyrecords.Selecting only timestampand anomaly.Appendinganomaly and timestamp.Partitioning trainingset and validation set.Random foresttraining.Validating.Node 24Node 26Converting predictedvalues to intin order to apply thescorer node. CSV Reader Line Plot Row Filter Row Filter CSV Reader 20% Math Formula queued Row Filter queued Table RowTo Variable queued Row Filter queued Row Filter queued Row Filter queued Column Filter queued Column Appender queued Partitioning queued Random Forest Learner(Regression) queued Random Forest Predictor(Regression) queued Numeric Scorer queued Scorer queued Double To Integer queued Visualization queued Anomaly detection Load and clean data from nicla sensor. Loading data from right arm. Reading the nicla.csv external sensorfile first, skipping the first 6 linesdue to faulty format/data.Visualizing data,we see the sensor idlingfor some time.By selecting column with |GyroX|>1we find that the sensor start workingfrom row 898, thus we ingore the first 897rows.We consider rows starting from row 898 to row 1728006to match right arm data.Node 10Resetting timestamps.Getting first row.Converting first timestamp to var.Selecting onlyanomlay states.Finding last occurenceof anomaly state.Cleaning last faultyrecords.Selecting only timestampand anomaly.Appendinganomaly and timestamp.Partitioning trainingset and validation set.Random foresttraining.Validating.Node 24Node 26Converting predictedvalues to intin order to apply thescorer node. CSV Reader Line Plot Row Filter Row Filter CSV Reader 20% Math Formula queued Row Filter queued Table RowTo Variable queued Row Filter queued Row Filter queued Row Filter queued Column Filter queued Column Appender queued Partitioning queued Random Forest Learner(Regression) queued Random Forest Predictor(Regression) queued Numeric Scorer queued Scorer queued Double To Integer queued Visualization queued

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: rikyv
Created at: 2023-08-28
On NodePit since: 2023-09-26
Last update: 2024-05-01
Created with KNIME version: v5.1.1
Tags: data analysisrobotics

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!