simulator or real life flag about carnd-capstone HOT 2 CLOSED

In the tl_detector package, there are two launch files. launch/tl_detector.launch is launched from the styx.launch file, and launch/tl_detector_site.launch is launched from site.launch. I think you should add a private parameter to tl_detector.py for the model file, e.g.

model_path = rospy.get_param("~model_path")

Then, set the model path differently in each of those launch files, e.g.

<?xml version="1.0"?>
<launch>
    <node pkg="tl_detector" type="tl_detector.py" name="tl_detector" output="screen" cwd="node">
        <param name="model_path" value="path/to/model"/>
    </node>
</launch>

from carnd-capstone.

This is implemented in tl_training

from carnd-capstone.