Awesome
Model Graph Extraction Difference
This is a HAROS plugin to calculate the Graph Extraction Difference of extracted models against a ground truth.
It should serve little to no purpose for regular HAROS users. It has been used to evaluate the performance of the HAROS model extractor in terms of Precision, Recall and F1-Score.
In order to use this plugin, you must pass it, as input in the project file, a ground truth of the expected Computation Graph. That is, you must list every detail of every entity (nodes, topics, services, parameters). For example:
%YAML 1.1
---
project: Fictibot
packages:
- fictibot_drivers
- fictibot_controller
- fictibot_multiplex
- fictibot_msgs
configurations:
multiplex:
launch:
- fictibot_controller/launch/multiplexer.launch
user_data:
haros_plugin_model_ged:
truth:
nodes:
/fictibase:
node_type: fictibot_drivers/fictibot_driver
traceability:
package: fictibot_controller
file: launch/multiplexer.launch
line: 2
column: 3
publishers:
- topic: /bumper
msg_type: std_msgs/Int8
queue_size: 21
traceability:
package: fictibot_drivers
file: src/sensor_manager.cpp
line: 10
column: 29
- topic: /laser
msg_type: std_msgs/Int8
queue_size: 21
traceability:
package: fictibot_drivers
file: src/sensor_manager.cpp
line: 11
column: 29
# ...
subscribers:
- topic: /stop_cmd
msg_type: std_msgs/Empty
queue_size: 21
traceability:
package: fictibot_drivers
file: src/motor_manager.cpp
line: 15
column: 30
- topic: /teleop_cmd
msg_type: std_msgs/Float64
queue_size: 21
traceability:
package: fictibot_drivers
file: src/motor_manager.cpp
line: 17
column: 30
# ...
Then, the plugin compares the automatically extracted model with this ground truth, and reports the final results, including incorrect, missing or spurious entities.