robokudo.descriptors.analysis_engines.tiago_demo

Analysis engine for TIAGo robot perception pipeline.

This module provides an analysis engine that demonstrates perception capabilities using the TIAGo robot’s camera system. It implements a query-based pipeline for tabletop segmentation and object pose estimation.

The pipeline implements the following functionality:

  • Query-based perception control

  • TIAGo camera data processing

  • Point cloud analysis and segmentation

  • Object pose estimation using PCA

  • Query result generation and response

Note

This engine is specifically designed for the TIAGo robot platform and uses its camera configuration. The pipeline can be extended with additional perception capabilities as needed.

Classes

AnalysisEngine

Analysis engine for TIAGo robot perception.

Module Contents

class robokudo.descriptors.analysis_engines.tiago_demo.AnalysisEngine

Bases: robokudo.analysis_engine.AnalysisEngineInterface

Analysis engine for TIAGo robot perception.

This class implements a pipeline that processes camera data from the TIAGo robot to perform tabletop segmentation and object pose estimation. It uses a query-based approach to control perception tasks.

The pipeline includes:

  • Query handling for perception control

  • TIAGo camera data collection and preprocessing

  • Point cloud analysis and segmentation

  • Object pose estimation using PCA

  • Query result generation and response

Note

The pipeline uses PCA-based pose estimation by default, but can be configured to use bounding box-based estimation by uncommenting the relevant annotator.

name()

Get the name of the analysis engine.

Returns:

The name identifier of this analysis engine

Return type:

str

implementation()

Create a pipeline for TIAGo robot perception.

This method constructs a processing pipeline that handles perception tasks for the TIAGo robot. The pipeline processes camera data to perform tabletop segmentation and object pose estimation.

Pipeline execution sequence:

  1. Initialize pipeline

  2. Wait for query

  3. Read TIAGo camera data

  4. Preprocess image

  5. Crop point cloud

  6. Detect table plane

  7. Extract object clusters

  8. Estimate object poses (PCA)

  9. Generate and send query response

Returns:

The configured pipeline for TIAGo perception

Return type:

robokudo.pipeline.Pipeline