Skip to content

Latest commit

 

History

History
185 lines (167 loc) · 5.85 KB

step3-make-racecar-interact-with-f1tenth-gym.md

File metadata and controls

185 lines (167 loc) · 5.85 KB
Raw

Subscribe to Simulation Data

  • Assuming you haven't changed the racecar_namespace:
  • For LaserScan data, you could subscribe to:
    • racecar1/scan
    • racecar2/scan
    • racecar3/scan
  • For Odometry data,:
    • racecar1/odom
    • racecar2/odom
    • racecar3/odom

Example Drive for Racecar

  • Below command sends an example AckermannDriveStamped message to the 1st racecar.
ros2 topic pub -r 100 racecar1/drive ackermann_msgs/msg/AckermannDriveStamped "{header: {stamp: {sec: 0, nanosec: 0}, frame_id: 'racecar1/base_link'}, drive: {steering_angle: 1.0, steering_angle_velocity: 1.0, speed: 1.0, acceleration: 0.0, jerk: 0.0}}"

Note:

  • While sending an AckermannDriveStamped message to a racecar, your AckermannDriveStamped message's frame_id should have the value of racecar_namespace_of_the_racecar_you_want_to_control/base_link. Consider the example above. You can observe that this message is being sent to racecar1/drive topic and within the message, it includes a frame_id of racecar1/base_link which should only cause racecar1 to move.

  • If you'd like to send an AckermannDriveStamped to another racecar, you'd need to adjust the topic you'll send the message to and the frame_id of your corresponding AckermannDriveStamped message.

Tune the parameters for the racing algorithm

Go to Racecar1 Reactive Gap Follower config for racecar1

Go to Racecar2 Reactive Gap Follower config for racecar2

/**/cass_gap_follower_slow_pace:
 ros__parameters:
  racecar_namespace: "racecar1"
  # Ranges Smoothing Filter Size
  # Dimension: 1 x (2*smoothing_filter_size-1)
  smoothing_filter_size: 1

  # Wall Follow PID controller gains

  # 2.45 radians 140 degrees (-70  >  +70)
  forward_view_angle: 2.0

  #rpilidar front index different than hokuyo
  scanner_forward_offsetangle: 0.0

  disparity_threshold: 0.3
  disparity_filter_size: 3
-  disparity_publish: true
+  disparity_publish: false
  car_width: 0.40

  # Desired Car Velocity based on Error
- error_based_ranges.low: 1.0
+ error_based_ranges.low: 2.0
  error_based_ranges.medium: 3.0
  error_based_ranges.high: 6.0
  error_based_velocities.low: 1.0
- error_based_velocities.medium: 1.5
+ error_based_velocities.medium: 1.7
  error_based_velocities.high: 2.0

  drive_topic: "drive"  # notice that namespaces differ for each racecar so you don't have to change drive and scan topics
  scan_topic: "scan"

After you tune the parameters,

cd ~/path_to_ws
docker compose up

to observe the results of your changes

What is a stack?

  • In the context of Eclipse Muto, a stack is a set of instructions written in a json-like structure to let Eclipse Muto know which ROS nodes it should introspect.

Example Stack

{
  "thingId": "org.eclipse.muto.sandbox:racecar01-gap-follower-fast-pace",
  "policyId": "org.eclipse.muto.sandbox:racecar01-gap-follower-fast-pace",
  "definition": "org.eclipse.muto:Stack:0.0.1",
  "attributes": {
    "type": "simulator"
  },
  "features": {
    "stack": {
      "properties": {
        "name": "Racecar1 Gap Follower (Fast Paced)",
        "context": "eteration_office",
        "stackId": "org.eclipse.muto.sandbox:racecar01-gap-follower-fast-pace",
        "arg": [],
        "node": [
          {
            "name": "cass_gap_follower_fast_pace",
            "pkg": "reactive_gap_follower",
            "exec": "cass_gap_follower",
            "param": [
              {
                "from": "$(find reactive_gap_follower)/config/params.yaml"
              }
            ]
          }
        ]
      }
    }
  }
}
  • If you pay attention to the "Node" part, it includes an instruction to start the reactive gap follower algorithm.
  • You can start your own nodes in a similar manner and provide a different parameter file.
  • Notice that while you're providing parameters, you could go like: 
    "param": [
  {
    "from": "$(find your_package_name)/path_to_your.yaml"
  }
]
  • or you could simply define your parameters in place like:
  "param": [
    {
      "param1": "value1",
      "param2": "value2"
    }
  ]

Pushing a stack to the twin server

If you have a stack like below:

{
  "thingId": "org.eclipse.muto.sandbox:racecar01-gap-follower-fast-pace",
  "policyId": "org.eclipse.muto.sandbox:racecar01-gap-follower-fast-pace",
  "definition": "org.eclipse.muto:Stack:0.0.1",
  "attributes": {
    "type": "simulator"
  },
  "features": {
    "stack": {
      "properties": {
        "name": "Racecar1 Gap Follower (Fast Paced)",
        "context": "eteration_office",
        "stackId": "org.eclipse.muto.sandbox:racecar01-gap-follower-fast-pace",
        "arg": [],
        "node": [
          {
            "name": "cass_gap_follower_fast_pace",
            "pkg": "reactive_gap_follower",
            "exec": "cass_gap_follower",
            "param": [
              {
                "from": "$(find reactive_gap_follower)/config/params.yaml"
              }
            ]
          }
        ]
      }
    }
  }
}

You can push it to the Twin Server using curl and introspect it using Muto Dashboard:

curl -X PUT -H "Content-Type: application/json" -d '
{
  "definition": "org.eclipse.muto:Stack:0.0.1",
  "attributes": {                              
    "type": "simulator"
  },                   
  "features": {
    "stack": {
      "properties": {
        "name": "Example Stack",
        "context": "eteration_office",
        "stackId": "org.eclipse.muto.sandbox:example-id",
        "arg": [],                                       
        "node": []
      }           
    }
  } 
}   
' https://ditto:[email protected]/api/2/things/org.eclipse.muto.sandbox:org.eclipse.muto.sandbox:example-id.launch

For further reading, refer to API Docs