Dataset

OpenCOOD is an Open COOperative Detection framework for autonomous driving. More details on the dataset can be found at OpenCOOD Github .

Dataset at LISHA's IoT

• domain: opencood
• username: opencood
• password: A7B64D415BD3E9AA
• version: 1.2 (Mobile)

Meta-Data

• Data-set type: "simulation"
• Simulator: "CARLA-0.9.15"
  • Map: "Town06"
  • Description: "Town 6 is a low density town set into a coniferous landscape exhibiting a multitude of large, 4-6 lane roads and special junctions like the Michigan Left."
  • random seed: 11 # seed for numpy and random
  • weather:
    • sun_altitude_angle: 15 # 90 is the midday and -90 is the midnight
    • cloudiness: 0 # 0 is the clean sky and 100 is the thickest cloud
    • precipitation: 0 # rain, 100 is the heaviest rain
    • precipitation_deposits: 0 # Determines the creation of puddles. Values range from 0 to 100, being 0 none at all and 100 a road completely capped with water.
    • wind_intensity: 0 # it will influence the rain
    • fog_density: 0 # fog thickness, 100 is the largest
    • fog_distance: 0 # Fog start distance. Values range from 0 to infinite.
    • fog_falloff: 0 # Density of the fog (as in specific mass) from 0 to infinity. The bigger the value, the more dense and heavy it will be, and the fog will reach smaller heights
    • wetness: 0
  • Map specificities:
    • Road modifications: [None]
    • Includes Traffic Lights: Yes
    • Includes Traffic Signals: Yes
  • Traffic Management:
    • Number of vehicles: 83
    • Range Background Vehicle Spawn:
      • x_max :470
      • y_max: 610
      • x_min: -30
      • y_min: -9
      • x_step: 10
      • y_step: 3.5

Vehicle Attributes

• Model: "vehicle.lincoln.mkz_2017"
• Color: '0, 0, 255' # green
• Extent:
  • length: 4.901684 # meters
  • width: 2.128324 # meters
  • height : 1.510746 # meters
• Torque Curve: # Curve that indicates the torque measured in Nm for a specific RPM of the vehicle's engine.

• Max RPM: 6500 # The maximum RPM of the vehicle's engine.
• moi: 1 # kg*m². The moment of inertia of the vehicle's engine.
• damping_rate_full_throttle: 0.05 # Damping ratio when the throttle is maximum.
• damping_rate_zero_throttle_clutch_engaged: 2 # Damping ratio when the throttle is zero with clutch engaged.
• damping_rate_zero_throttle_clutch_disengaged: 0.35 # Damping ratio when the throttle is zero with clutch disengaged.
• use_gear_autobox: True # If True, the vehicle will have an automatic transmission.
• gear_switch_time: 0.1 # Switching time between gears.
• clutch_strength: 10 # kg*m²/s Clutch strength of the vehicle.
• final_ratio: 3.21 # Fixed ratio from transmission to wheels.
• forward_gears: # Down = Quotient between current RPM and MaxRPM where the autonomous gear box should shift down. Up = Quotient between current RPM and MaxRPM where the autonomous gear box should shift up.

• Mass: 1696.0 # Mass of the vehicle in kg
• drag_coefficient: 0.300000 # Drag coefficient of the vehicle's chassis.
• center_of_mass: [0.600000, 0.000000, -0.300000] # Center of mass of the vehicle.
• steering_curve: # Curve that indicates the maximum steering for a specific forward speed.

• wheels:
  • tire_friction: 3.500000 # A scalar value that indicates the friction of the wheel.
  • damping_rate: 0.250000 # Damping rate of the wheel.
  • max_steer_angle: 69.999992 (front), 0 (back) # Maximum angle that the wheel can steer in degrees.
  • radius: 35.500000 # Radius of the wheel in centimeters
  • max_brake_torque: 800.000000 # Maximum brake torque in N*m
  • max_handbrake_torque: 0.000000 (front), 1600.000000 (back) # Maximum handbrake torque in N*m
  • lat_stiff_max_load: 3.000000 # Maximum normalized tire load at which the tire can deliver no more lateral stiffness no matter how much extra load is applied to the tire. Each vehicle has a custom value.
  • lat_stiff_value: 20.000000 # Maximum stiffness per unit of lateral slip. Each vehicle has a custom value.
  • long_stiff_value: 3000.000000 # kg/rad. Tire longitudinal stiffness per unit gravitational acceleration. Each vehicle has a custom value.
  • Wheel position: # World position of the wheel.
  • front left wheel: [10872.801758, 5181.030762, 65.987663]
  • front right wheel: [11029.766602, 5182.101074, 65.987663]
  • back left wheel: [10870.842773, 5468.136719, 65.987663]
  • back right wheel: [11027.808594, 5469.207520, 65.987663]

Vehicle Driving Agent

• Controller: PID
  • lat: k_p: 0.75, k_d: 0.02, k_i: 0.4
  • lon: k_p: 0.37, k_d: 0.024, k_i: 0.032
  • max_brake : 1.0
  • max_throttle: 1.0
  • max_steering: 0.3
• Agent: "carla.agents.BehaviorAgent"
  • Collision_time_ahead: 1.5 # used for collision checking
  • emergency_param: 0.4 # used to identify whether a emergency stop needed
  • ignore_traffic_light: true
  • local_planner:
    • buffer_size: 12 # waypoint buffer size
    • min_dist: 3 # used to pop out the waypoints too close to current location
    • trajectory_dt: 0.2 # for every dt seconds, we sample a trajectory point from the trajectory
    • trajectory_update_freq: 15 # used to control trajectory points updating frequency
    • waypoint_update_freq: 9 # used to control waypoint updating frequency
  • max_speed: 41 # The maximum speed in km/h your vehicle will be able to reach.
  • overtake_allowed: false
  • overtake_counter_recover: 35
  • safety_time: 4 # Time-to-collision; an approximation of the time it will take for your vehicle to collide with one in front if it brakes suddenly.
  • sample_resolution: 4.5 # the unit distance between two adjacent waypoints in meter
  • speed_decrease: 15 # How quickly in km/h your vehicle will slow down when approaching a slower vehicle ahead.
  • speed_lim_dist: 3 # Value in km/h that defines how far your vehicle's target speed will be from the current speed limit (e.g., if the speed limit is 30km/h and speed_lim_dist is 10km/h, then the target speed will be 20km/h)
  • tailgate_speed: 51 # when a vehicles needs to be close to another vehicle asap

Sensor Attributes

• IMU:
  • Type: "carla.sensor.other.imu"
  • Location: [0,0,1] # meters, relative to vehicle center of mass

• GNSS:
  • Type: "carla.sensor.other.gnss"
  • Description: calculated by adding the metric position to an initial geo reference location defined within the OpenDRIVE map definition
  • Location: [0,0,1] # meters, relative to vehicle center of mass

• Camera:
  • Type: "Carla.RGB"
  • Description: "Frontal Camera"
  • Lens: "carla.lens" # configuration of lens is given by blueprint attribute
  • Location: [2.5, 0, 1] # meters, relative to vehicle center of mass
  • Rotation: [Pitch : -35, Yaw : 0, Roll: 0] # Degrees, from -180 to 180

• LIDAR:
  • Type: "carla.sensor.lidar.ray_cast"
  • Description: "Frontal LIDAR. sensor simulates a rotating LIDAR implemented using ray-casting. The points are computed by adding a laser for each channel distributed in the vertical FOV. The rotation is simulated computing the horizontal angle that the Lidar rotated in a frame. The point cloud is calculated by doing a ray-cast for each laser in every step. points_per_channel_each_step = points_per_second / (FPS * channels). A LIDAR measurement contains a package with all the points generated during a 1/FPS interval. During this interval the physics are not updated so all the points in a measurement reflect the same "static picture" of the scene."
  • Location: [2.5, 0, 1] # meters, relative to vehicle center of mass
  • Rotation: [Pitch : -35, Yaw : 0, Roll: 0] # Degrees, from -180 to 180

Simulations

There are two simulations available with multiple vehicles. Each vehicle operates as EGO and collects frontal camera and lidar samples as described further below. Each vehicle is accompanied by a collection of vehicle ground truths. The vehicles are selected at each timestamp whenever they are hit by more than 3 lidar rays. This means that the time series of each ground truth has a different size. If a vehicle does not have a sample at a specific timestamp it could not be identified through perception.

• Simulation 1:
  • Vehicles:
    • signature: 641 , t0: 1629145617450000 , tf: 1629145659250000 , length= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 650, 659, 669, 673, 677, 682, 684, 685, 689, 690, 691, 692, 693, 695, 696, 698, 706, 708, 710, 712, 714, 720, 727, 728, 729, 730, 732, 740, 744, 746, 747, 748, 751, 688, 679, 750, 671, 737
    • signature: 650 , t0: 1629145617450000 , tf: 1629145659250000 , length= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 641, 659, 671, 677, 682, 684, 685, 687, 688, 689, 691, 692, 693, 698, 705, 708, 710, 712, 714, 720, 727, 728, 730, 732, 737, 740, 744, 746, 747, 748, 750, 751, 695, 669, 676, 738, 755, 706
    • signature: 659 , t0: 1629145617450000 , tf: 1629145659250000 , length= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 641, 650, 671, 677, 685, 687, 689, 691, 692, 693, 695, 698, 705, 706, 708, 712, 714, 727, 730, 732, 737, 738, 740, 746, 747, 748, 750, 751, 755, 688, 676, 669, 733, 739, 743, 718, 682, 717, 736, 713, 683, 742, 710

• Simulation 2:
  • Vehicles:
    • signature: 1188 , t0: 1629306665500000 , tf: 1629306677600000 , len= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 1197, 1206, 1225, 1227, 1228, 1229, 1231, 1226, 1232, 1230, 1215, 1233
    • signature: 1197 , t0: 1629306665500000 , tf: 1629306677600000 , len= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 1188, 1206, 1225, 1229, 1231, 1227, 1228, 1226
    • signature: 1206 , t0: 1629306665500000 , tf: 1629306677600000 , len= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 1188, 1197, 1215, 1225, 1227, 1228, 1229, 1231, 1226, 1232, 1230, 1233
    • signature: 1215 , t0: 1629306665500000 , tf: 1629306677600000 , len= 4.901683330535889 , width= 2.128324270248413 , height= 1.5107464790344238
      • Ground Truth of Perception: Signatures: 1206, 1226, 1227, 1228, 1230, 1232, 1233, 1224, 1188

Vehicle Classes

Vehicle Motion Vectors are stored with a unit containing the class of the vehicle (ETSI). The class represents is related to the dimensions of the vehicle:

The ground truth motion vectors have all been stored as Passenger Cars.

IMU Data

• Period: 100ms

Control Data

• Period: 100ms

LIDAR Data

• Period: 100ms

• Original format: ".pcd" with x, y, z, and color as columns (values in F32)
• Convert blob returned query back to PCD: base64_decode twice.

Camera Data

• Period: 100ms

• Original format PNG 4096x4096 pixels
• Conversion: None, image is read as binary and sent to platform as a blob
• Convert blob returned query back to PNG: base64_decode twice.

Getters

A set of getter examples in multiple different languages is available in the IoT Platform documentation - Example Scripts.
Important Note: On the retrieval of large data (digital data with size > 8 bytes e.g., camera, lidar or MV Global), a single base64 decodes is required to recover the original sample.
Note that a mobile series is represented with at least:

"series" : Object {
    "unit" : unsigned long
    "t0" : unsigned long long
    "tf" : unsigned long long
    "signature" : unsigned int
}

Note that the Global Motion Vectors from the ground truth can be read with the following Python code:

speed, heading, yawrate, acceleration= struct.unpack("ffff",value)

Construction and Usage of CARLA Vehicle Telemetry Dashboard in Grafana

This Grafana dashboard is designed to provide a structured visualization of vehicle telemetry and motion parameters within the CARLA simulator, utilizing the OpenCood SmartData Model (https://lisha.ufsc.br/Opencood+-+SmartData+Model). By organizing data within this model, the dashboard enhances the monitoring of vehicle dynamics and environmental interactions in real time.

Description of Variables and Data Collection

The dashboard visualizes several key parameters, which are structured as follows:

IMU Data: The IMU acceleration and rates are essential for understanding vehicle motion. The IMU acceleration graph tracks three-dimensional acceleration—longitudinal, lateral, and vertical—which are crucial to gauge how the vehicle accelerates and decelerates relative to its axes. Meanwhile, the yaw, roll, and pitch rates measure rotational movements, indicating changes in orientation that are critical for assessing stability and control.

Telemetry Data: This section includes variables such as engine RPM, gear, longitudinal speed, and drag force. The drag graph displays the force exerted on the vehicle due to air resistance, measured in Newtons, offering insights into how vehicle speed and environmental factors impact fuel efficiency and performance. The telemetry graph consolidates engine RPM, longitudinal speed, vertical speed, and gear into a cohesive visualization, allowing for simultaneous tracking of the vehicle's speed, power output, and transmission state.

GNSS Data: Geographic data is represented here, including latitude, longitude, and altitude. These GNSS parameters help with precise location tracking within the simulation, providing spatial context necessary for navigation and route analysis.

Structure of the Grafana Dashboard

Following the principles of SmartData’s Series Semantics (https://epos.lisha.ufsc.br/Series+Semantics), the dashboard organizes and filters data based on predefined regions of interest. In version 1.2 of SmartData, this region is defined as a spherical area with a center at “(x, y, z)” and a radius ”r”. By limiting data to this spherical region, the dashboard improves data retrieval efficiency, ensuring only relevant telemetry data is displayed.

For example, data on drag force is only visualized when within the region of interest, which helps to streamline the analysis by focusing on areas where drag is most likely to affect vehicle dynamics.

Disambiguation and Signature

SmartData uses a "dev" identifier as a disambiguator for variables that may share similar units but represent different phenomena. For example, drag force and other forces are measured in Newtons but serve different purposes in telemetry analysis; the "dev" label allows Grafana to distinguish between them without ambiguity. This ensures that each graph on the dashboard clearly represents distinct data types, enhancing interpretability for simulation analysts.

Additionally, each data entry has a unique signature associated with the specific vehicle in the simulation. This signature serves as a unique identifier, allowing data tracking across sessions and enabling a reliable historical analysis of vehicle behavior.

Workflow Process

The workflow in this dashboard setup involves transforming raw telemetry data from CARLA into the structured visualizations seen on Grafana. Data is initially queried from the SmartData model, then processed and mapped to fit the dashboard’s layout. Each graph uses processed data to render real-time visuals based on axis scaling, color schemes, and other parameters tailored for clarity and consistency.

For instance, the IMU Rates graph displays yaw, roll, and pitch rates based on continuously updated IMU values. This setup allows for a clear depiction of the vehicle's rotational dynamics, providing real-time insights into its orientation and stability. Similarly, the GNSS graph filters the vehicle’s location data to display coordinates within the specified "region of interest," ensuring the focus remains on relevant spatial data.

Summary

This CARLA telemetry dashboard, supported by SmartData version 1.2, offers a structured, efficient approach to monitoring vehicle performance in real time. Using SmartData features like the "dev" disambiguator and region of interest filtering, the dashboard delivers accurate, organized data essential for assessing and optimizing vehicle behavior within simulation scenarios. This setup provides a valuable resource for technical analysis and performance tuning in CARLA-based testing.