Skip to main content

Searching for and visualizing logs

In this example scenario, your organization's robot operators and technicians have noticed an increase in optical encoder failures over time in the fleet and asked for help with investigating the root cause. They opened a ticket providing details on the robot requiring replacement.

Robot ID: ROBO3396
Date of failure: 09/03/2024
Date of replacement: 09/04/2024

You would like to analyze data from the last run of the robot before the failure.

To search for logs from that specific robot, begin by searching by Robot ID.

  1. Navigate to the Search page by clicking on Search in the left navigation menu.

Search Nav

  1. Click on the Comparative search tab.

Comparative Search

  1. In the Conditions section, click on the Select rule dropdown and select Robot ID.

  2. Select the Equal to condition and type or paste ROBO3396 into the filter value field.

  3. To narrow down the search to the date of failure, add a date rule. Click + Add rule.

Add Rule

  1. Click on theSelect Rule dropdown and select Start time.

  2. Select the Greater than condition and select 09/03/2024 in the date picker. It will default to 09/03/2024 12:00:00 AM.

  3. Click on the Search button. In the results pane below, you will see all of the logs generated by robot ROBO3396 on 09/03/2024 or later. You should see three logs.

NOTE: The ROBOLOG IDs you see will be different from the screenshots provided, as they are unique for your demo organization. However, the robolog names will be the same.

  1. Click on the log name /generated_20240903_190944.bag. This will take you to the Metadata Details page for that log.

Click on Log Name

  1. On the Metadata Details page, you can see the metadata attributes for the log, including the Model-Prime Robolog ID, Robolog URI, Duration, and channel information. You can also see the status of Model-Prime processes such as Metadata Ingestion, which populates the metadata attributes, and Introspective Ingestion, which prepares log message data for analysis.

Metadata Details

Knowing that temperature may impact performance of optical encoders, you can explore the temperature from the robot's last run prior to the failure by visualizing the channel data in this log.

  1. Click on the Visualization tab.

  2. In the Log visualization pane, click on + Add channel

  3. Select the sensor_msgs/Temperature Channel Type. The /terrible_temperature Channel Name will populate. Select the temperature data path.

You will see the temperature plot appear in the graph below.

Metadata Log Viz

The plot shows some temperature spikes that warrant further investigation. To explore and analyze data across multiple logs from this robot, you can use a SageMaker Notebook that runs against the Model-Prime Analytics Platform.