Setup

This section describes the settings required to use driving_log_replayer.

Cli setting

In order to reduce the number of arguments to be passed to driving_log_replayer_cli, the directories specified as arguments are described in a configuration file.

Therefore, before using the cli, create a $HOME/.driving_log_replayer.config.toml file in the format presented below. You can create it manually or by using the dlr configuration command.

# manual creation
nano $HOME/.driving_log_replayer.config.toml

# cli creation
dlr configuration register -d ${data_directory} -o ${output_directory} -a ${autoware_path} [-p ${profile}]

At least one profile is required, and one must be named default.

Specifying the profile name with -p ${profile_name} in the commands described below will load the settings specified in the profile. You can switch between multiple autoware profiles, and if no profile is specified, default is used.

[profile_name]
data_directory = "path of the data folder used as input for the simulation"
output_directory = "path of the folder to output the simulation results"
autoware_path = "path of the autoware workspace folder"

example setting

# default is required and selected when profile name is omitted
[default]
data_directory = "$HOME/driving_log_replayer_data/default"
output_directory = "$HOME/driving_log_replayer_output/default"
autoware_path = "$HOME/autoware"

[localization]
data_directory = "$HOME/driving_log_replayer_data/localization"
output_directory = "$HOME/driving_log_replayer_output/localization"
autoware_path = "$HOME/autoware"

[yabloc]
data_directory = "$HOME/driving_log_replayer_data/yabloc"
output_directory = "$HOME/driving_log_replayer_output/yabloc"
autoware_path = "$HOME/autoware"

[ar_tag_based_localizer]
data_directory = "$HOME/driving_log_replayer_data/ar_tag_based_localizer"
output_directory = "$HOME/driving_log_replayer_output/ar_tag_based_localizer"
autoware_path = "$HOME/autoware"

[obstacle_segmentation]
data_directory = "$HOME/driving_log_replayer_data/obstacle_segmentation"
output_directory = "$HOME/driving_log_replayer_output/obstacle_segmentation"
autoware_path = "$HOME/autoware"

[perception]
data_directory = "$HOME/driving_log_replayer_data/perception"
output_directory = "$HOME/driving_log_replayer_output/perception"
autoware_path = "$HOME/autoware"

Folder structure and file naming rules

This section describes the folder structure and file naming rules expected by driving_log_replayer.

In driving_log_replayer the folder structure and file names are fixed to reduce the number of paths to be described in the scenario and the arguments to be passed to the command. Also, by placing multiple folders in data_directory multiple tests can be executed in succession.

Data Folder

A folder where resources used in the simulation are stored.

For each test case: a scenario, rosbag, and dataset are placed.

Data Folder Structure for use case without t4_dataset

profile_data_directory                // .driving_log_replayer.config の data_directory
│
├── TC001                          // Test case directory. Directry name can be named arbitrarily
│   ├── scenario.yaml             // Scenario
│   └── input_bag                 // Bag for input containing sensor data
│       ├── input_bag_0.db3       // Binary file of bag
│       └── metadata.yaml         // Metadata file of bag
│
├── TC002                          // Test case directory. Same structure as TC001
...

Data Folder Structure for use case with t4_dataset

profile_data_directory                 // .driving_log_replayer.config の data_directory
│
├── TC001                           // Test case directory. Directry name can be named arbitrarily
│   ├── scenario.yaml              // Scenario
│   └── t4_dataset
│       ├── T4D001                 // t4_dataset directory. If use case is sensing, t4_dataset is always one.
│       │   ├── annotation
│       │   ├── data
│       │   │   └── LIDAR_CONCAT
│       │   └── input_bag
│       └── T4D002                 // t4_dataset directory. If use case is peception, t4_dataset can be multiple.
│           ├── annotation
│           ├── data
│           │   └── LIDAR_CONCAT
│           └── input_bag
│          ...
│
├── TC002                           // Test case directory. Same structure as TC001
...

When you want to use a common scenario file in a profile

This is used when you want to evaluate bag files acquired with the same map, vehicle, and sensor configuration in one common scenario. It can be used when you want to run simulations that only require metrics to determine threshold values with annotationless_perception.

For example, in the following case, TC001 is evaluated in base_scenario.yaml and TC002 uses scenario.yaml in the directory

profile_data_directory                // .driving_log_replayer.config の data_directory
├── base_scenario.yaml　　　　　　 // Common scenario used when no scenario file exists in the directory
│
├── TC001                          // Test case directory. Directry name can be named arbitrarily
│   └── input_bag                 // Bag for input containing sensor data
│       ├── input_bag_0.db3       // Binary file of bag
│       └── metadata.yaml         // Metadata file of bag
│
├── TC002                          // Test case directory. Directry name can be named arbitrarily
│   ├── scenario.yaml             // The scenario files in the directory are used. base_scenario.yaml is ignored
│   └── input_bag                 // Bag for input containing sensor data
│       ├── input_bag_0.db3       // Binary file of bag
│       └── metadata.yaml         // Metadata file of bag

Map Folder

A folder where all the maps used in the simulation are stored.

autoware_map
│
├── LocalMapPath1            // Path specified by LocalMapPath in the scenario
│   ├── lanelet2_map.osm    // lanelet file
│   └── pointcloud_map.pcd  // pcd file
│
├── LocalMapPath2            // Path specified by LocalMapPath in the scenario
...