Merge pull request #99 from motional/update-devkit-v0.4

Update devkit to v0.4

BUILD

@@ -3,9 +3,10 @@ load("@com_github_bazelbuild_buildtools//buildifier:def.bzl", "buildifier")
 package(default_visibility = ["//visibility:public"])
 
 filegroup(
-    name = "jenkins_env",
+    name = "requirements",
     data = [
         "requirements.txt",
+        "requirements_torch.txt",
     ],
 )
 

README.md

@@ -15,7 +15,7 @@ ______________________________________________________________________
   <a href="https://github.com/motional/nuplan-devkit/blob/master/tutorials/nuplan_framework.ipynb">Tutorial</a>
 </p>
 
-[![python](https://img.shields.io/badge/python-3.8%20%7C%203.9-blue.svg)]()
+[![python](https://img.shields.io/badge/python-%20%203.9-blue.svg)]()
 [![license](https://img.shields.io/badge/license-Apache%202.0-blue.svg)](https://github.com/motional/nuplan-devkit/blob/master/LICENSE.txt)
 
 ______________________________________________________________________
@@ -32,8 +32,10 @@ Stay tuned for the upcoming nuPlan ML planning challenges!
 ______________________________________________________________________
 
 ## Changelog
+- Aug 05 2022
+  * v0.4 Devkit: Database optimization, PYGEOS warning suppression, metrics improvements, scenario filter for training.
 - Jul 15 2022
-  * v0.3 Devkit: Updates to the devit including but not limited to: nuBoard update, reduce RAM usage during caching and training, new map APIs, simulation and metrics runtime improvements. 
+  * v0.3 Devkit: Updates to the devit including but not limited to: nuBoard update, reduce RAM usage during caching and training, new map APIs, simulation and metrics runtime improvements.
 - Jun 10 2022
   * v1.0 Dataset: Full nuPlan dataset release with over 1,300 hours of driving data (15,000+ logs) across 4 cities (Las Vegas, Pittsburgh, Boston, Singapore).
   * v0.2 Devkit: Multiple devkit updates with improvements across the whole framework (planning models, training, simulation, metrics, dashboard, tutorials and more).

nuplan/cli/BUILD

@@ -6,8 +6,8 @@ py_library(
     name = "db_cli",
     srcs = ["db_cli.py"],
     deps = [
-        "//nuplan/database/nuplan_db:nuplandb",
-        "//nuplan/database/nuplan_db:scenario_tag",
+        "//nuplan/database/nuplan_db:db_cli_queries",
+        "//nuplan/planning/scenario_builder/nuplan_db:nuplan_scenario_utils",
     ],
 )
 

nuplan/cli/db_cli.py

@@ -3,15 +3,40 @@
 
 import typer
 
-from nuplan.database.nuplan_db.nuplandb import NuPlanDB
-from nuplan.database.nuplan_db.scenario_tag import ScenarioTag
+from nuplan.database.nuplan_db.db_cli_queries import (
+    get_db_description,
+    get_db_duration_in_us,
+    get_db_log_duration,
+    get_db_log_vehicles,
+    get_db_scenario_info,
+)
+from nuplan.planning.scenario_builder.nuplan_db.nuplan_scenario_utils import download_file_if_necessary
 
 cli = typer.Typer()
 
 NUPLAN_DATA_ROOT = os.getenv('NUPLAN_DATA_ROOT', "/data/sets/nuplan/")
 NUPLAN_DB_VERSION = f'{NUPLAN_DATA_ROOT}/nuplan-v1.0/mini/2021.07.16.20.45.29_veh-35_01095_01486.db'
 
 
+def _ensure_file_downloaded(data_root: str, potentially_remote_path: str) -> str:
+    """
+    Attempts to download the DB file from a remote URL if it does not exist locally.
+    If the download fails, an error will be raised.
+    :param data_root: The location to download the file, if necessary.
+    :param potentially_remote_path: The path to the file.
+    :return: The resulting file path. Will be one of a few options:
+        * If potentially_remote_path points to a local file, will return potentially_remote_path
+        * If potentially_remote_file points to a remote file, it does not exist currently, and the file can be successfully downloaded, it will return the path of the downloaded file.
+        * In all other cases, an error will be raised.
+    """
+    output_file_path: str = download_file_if_necessary(data_root, potentially_remote_path)
+
+    if not os.path.exists(output_file_path):
+        raise ValueError(f"{potentially_remote_path} could not be downloaded.")
+
+    return output_file_path
+
+
 @cli.command()
 def info(
     db_version: str = typer.Argument(NUPLAN_DB_VERSION, help="The database version."),
@@ -20,11 +45,24 @@ def info(
     """
     Print out detailed information about the selected database.
     """
-    # Construct database
-    db = NuPlanDB(load_path=db_version, data_root=data_root)
-    # Use the default __str__
-    typer.echo("DB info")
-    typer.echo(db)
+    db_version = _ensure_file_downloaded(data_root, db_version)
+    db_description = get_db_description(db_version)
+
+    for table_name, table_description in db_description.tables.items():
+        typer.echo(f"Table {table_name}: {table_description.row_count} rows")
+
+        for column_name, column_description in table_description.columns.items():
+            typer.echo(
+                "".join(
+                    [
+                        f"\tcolumn {column_name}: {column_description.data_type} ",
+                        "NULL " if column_description.nullable else "NOT NULL ",
+                        "PRIMARY KEY " if column_description.is_primary_key else "",
+                    ]
+                )
+            )
+
+        typer.echo()
 
 
 @cli.command()
@@ -35,16 +73,11 @@ def duration(
     """
     Print out the duration of the selected db.
     """
-    # Construct database
-    db = NuPlanDB(load_path=db_version, data_root=data_root)
-
-    # Approximate the duration of db by dividing the number of lidar_pc and the frequency of the DB
-    assumed_db_frequency = 20
-    db_duration_s = len(db.lidar_pc) / assumed_db_frequency
+    db_version = _ensure_file_downloaded(data_root, db_version)
+    db_duration_us = get_db_duration_in_us(db_version)
+    db_duration_s = float(db_duration_us) / 1e6
     db_duration_str = time.strftime("%H:%M:%S", time.gmtime(db_duration_s))
-    typer.echo(
-        f"DB approximate duration (assuming db frequency {assumed_db_frequency}Hz) is {db_duration_str} [HH:MM:SS]"
-    )
+    typer.echo(f"DB duration is {db_duration_str} [HH:MM:SS]")
 
 
 @cli.command()
@@ -55,20 +88,15 @@ def log_duration(
     """
     Print out the duration of every log in the selected db.
     """
-    # Construct database
-    db = NuPlanDB(load_path=db_version, data_root=data_root)
-
-    # Approximate the duration of db by dividing the number of lidar_pc and the frequency of the DB
-    assumed_db_frequency = 20
+    db_version = _ensure_file_downloaded(data_root, db_version)
+    num_logs = 0
+    for log_file_name, log_file_duration_us in get_db_log_duration(db_version):
+        log_file_duration_s = float(log_file_duration_us) / 1e6
+        log_file_duration_str = time.strftime("%H:%M:%S", time.gmtime(log_file_duration_s))
+        typer.echo(f"The duration of log {log_file_name} is {log_file_duration_str} [HH:MM:SS]")
+        num_logs += 1
 
-    # Print out for every log the approximate durations
-    typer.echo(f"The DB: {db.name} contains {len(db.log)} logs")
-
-    for log in db.log:
-        lidar_pcs = [lidar for scene in log.scenes for lidar in scene.lidar_pcs]
-        db_duration_s = len(lidar_pcs) / assumed_db_frequency
-        db_duration_str = time.strftime("%H:%M:%S", time.gmtime(db_duration_s))
-        typer.echo(f"\tThe approximate duration of log {log.logfile} is {db_duration_str} [HH:MM:SS]")
+    typer.echo(f"There are {num_logs} total logs.")
 
 
 @cli.command()
@@ -79,14 +107,9 @@ def log_vehicle(
     """
     Print out vehicle information from every log in the selected database.
     """
-    # Construct database
-    db = NuPlanDB(load_path=db_version, data_root=data_root)
-
-    # Print out for every log the used vehicle
-    typer.echo("The used vehicles for every log follow:")
-
-    for log in db.log:
-        typer.echo(f"\tFor the log {log.logfile} vehicle {log.vehicle_name} of type {log.vehicle_type} was used")
+    db_version = _ensure_file_downloaded(data_root, db_version)
+    for log_file, vehicle_name in get_db_log_vehicles(db_version):
+        typer.echo(f"For the log {log_file}, vehicle {vehicle_name} was used.")
 
 
 @cli.command()
@@ -97,21 +120,13 @@ def scenarios(
     """
     Print out the available scenarios in the selected db.
     """
-    # Construct database
-    db = NuPlanDB(load_path=db_version, data_root=data_root)
-
-    # Read all available tags:
-    available_types = [tag[0] for tag in db.session.query(ScenarioTag.type).distinct().all()]
-
-    # Tag table
-    tag_table = db.scenario_tag
-
-    # Print out the available scenarios
-    typer.echo(f"The available scenario tags from db: {db_version} follow, in total {len(available_types)} scenarios")
+    db_version = _ensure_file_downloaded(data_root, db_version)
+    total_count = 0
+    for tag, num_scenarios in get_db_scenario_info(db_version):
+        typer.echo(f"{tag}: {num_scenarios} scenarios.")
+        total_count += num_scenarios
 
-    for tag in available_types:
-        tags = tag_table.select_many(type=tag)
-        typer.echo(f"\t - {tag} has {len(tags)} scenarios")
+    typer.echo(f"TOTAL: {total_count} scenarios.")
 
 
 if __name__ == '__main__':

nuplan/cli/test/BUILD

@@ -8,7 +8,6 @@ py_test(
     srcs = ["test_nuplan_cli.py"],
     deps = [
         "//nuplan/cli:nuplan_cli",
-        "//nuplan/common/actor_state:oriented_box",
-        "//nuplan/common/actor_state:state_representation",
+        "//nuplan/database/nuplan_db:db_description_types",
     ],
 )