本文へ移動
cccskills
無料GitHub で公開

record-simulation

Records a running CARLA simulation to a .log file via start_recorder/stop_recorder for later deterministic replay — capturing actor spawns, transforms, traffic-light states and animations (optionally velocities/controls). Use when the user asks to "record the simulation", "start/stop the recorder", "capture this run", or "save a scenario to replay later". Does NOT capture sensor images.

インストール方法を見る

含まれるファイル(5)

  • SKILL.md4.5 KB
  • references/recorder.md1.9 KB
  • scripts/check_env.sh2.1 KB
  • scripts/env.sh1.3 KB
  • scripts/record.py6.0 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Record a CARLA simulation

Paths. scripts/… and references/… below are relative to the directory holding this SKILL.md. Your working directory is the user's project, not that directory, so prefix them with its absolute path or the command is not found.

Capture a running simulation to a .log you can replay later (replay-recording) or interrogate (query-recording). The deliverable is a verified log: after stopping, the file is parsed back from the server to confirm it holds frames — start/stop alone confirm nothing.

What it captures: world state per frame — actor create/destroy, transforms, traffic-light states, vehicle wheel/animation and walker bones. With --extra (additional_data) also velocities, accelerations and control inputs. Not captured: sensor output (camera/lidar) — those re-simulate only on replay.

The .log is written on the server's filesystem (see the reference for exactly where).

Instructions

Progress:
- [ ] Step 1: Check prerequisites (bash scripts/check_env.sh), clear FAILs
- [ ] Step 2: Start recording (or use `clip` for a fixed duration)
- [ ] Step 3: Let the simulation run (traffic, scenario, manual drive, ...)
- [ ] Step 4: Stop, then verify the log parses back with frames

Commands need CARLA_HOST/CARLA_PORT from scripts/env.sh. Prefix with source scripts/env.sh or export them.

Step 1: Check prerequisites

bash scripts/check_env.sh

Step 2-4: Record

source scripts/env.sh

# manual bracketing: start, do things, stop
python3 scripts/record.py start --file /tmp/run.log
# ... run traffic / drive / trigger a scenario ...
python3 scripts/record.py stop

# or record a fixed-length clip and auto-verify
python3 scripts/record.py clip --file /tmp/run.log --seconds 20

# capture velocities/controls too (larger file)
python3 scripts/record.py start --file /tmp/run.log --extra

Use an absolute --file so you know where the log lands (a relative name goes under the server's CarlaUE4/Saved/).

Verify

clip verifies automatically; after a manual stop, confirm with the query-recording skill:

python3 ../query-recording/scripts/query.py info --file /tmp/run.log

A valid log shows a frame count and a duration. Zero frames means nothing ran between start and stop, or the path was wrong.

Examples

Example 1: capture a traffic run

User says: "record 30 seconds of traffic"

Start traffic (traffic-manager skill), then clip --file /tmp/traffic.log --seconds 30. VERIFY shows frames + duration ~30s.

Example 2: record around a manual action

User says: "record while I run my scenario"

start --file /tmp/scenario.log, run the scenario, stop, then query info.

Troubleshooting

Problem: log has 0 frames / query says empty Cause: nothing simulated between start and stop, or a bad server-side path. Solution: ensure actors are moving; use an absolute writable server path.

Problem: replayed run has no camera images Cause: the recorder never stores sensor data. Solution: expected — re-attach sensors and replay with --replay-sensors (replay-recording skill) to regenerate them.

Problem: file not found where I expected Cause: relative name resolves on the server, under CarlaUE4/Saved/. Solution: pass an absolute path, and remember it is on the server machine.

Outputs

Detail (capture contents, file location, size) in references/recorder.md.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Adds a new ROS 2 message type to CARLA's native ROS 2 interface — the hand-written POD struct under LibCarla/source/carla/ros2/types/msg, its Fast-CDR serialize/deserialize pair, the RIHS01 type hash computed in Docker, the CdrTopicInfo registration and the type-hash test. Use when the user asks to "add a ROS message type", "support sensor_msgs/X in CARLA", "add a carla_msgs type", "why is my type hash warning appearing", or needs a message CARLA does not publish yet. There is no IDL codegen — every step is manual.

日本語の概要は準備中です。原文の説明を表示しています。

carla-simulator/carla-agentic-tools42026年9月12日 更新

Adds a new topic to CARLA's native ROS 2 interface — a publisher (BasePublisher + PublisherImpl, topic suffix, QoS, registration in ROS2::GetOrCreateSensor, a ProcessDataFrom* entry point and its call site in the UE4 plugin) or a subscriber that lets ROS drive an actor. Also covers adding a middleware (RMW) behind IPublisherMiddleware. Use when the user asks to "publish X to ROS from CARLA", "add a ROS topic/publisher/subscriber", "make the obstacle/lane-invasion sensor publish", "add odometry to ROS", or "add another RMW".

日本語の概要は準備中です。原文の説明を表示しています。

carla-simulator/carla-agentic-tools42026年9月12日 更新

Reads and summarises ScenarioRunner output — the criteria pass/fail tables from --output/--file, the machine-readable --json and --junit result files, and the criteria JSON written alongside a --record recording — and runs the metrics module (metrics_manager.py) to compute custom measurements over a recorded run offline, without the simulator. Use when the user asks "did the scenario pass", "why did it fail", "summarise these results", "compare these runs", or wants distance/speed/lane metrics from a recording.

日本語の概要は準備中です。原文の説明を表示しています。

carla-simulator/carla-agentic-tools42026年9月12日 更新

Gets actor and level 3D bounding boxes and projects them into a camera image as 2D boxes — the dataset-annotation workflow. Lists box geometry, draws 3D boxes in the world via debug, or captures one camera frame and writes an annotated PNG plus a JSON of 2D boxes for matching actors. Use when the user asks to "get bounding boxes", "draw boxes around the cars", "project boxes onto the camera", or "generate a detection dataset / annotations".

日本語の概要は準備中です。原文の説明を表示しています。

carla-simulator/carla-agentic-tools42026年9月12日 更新

Builds CARLA (branch ue4-dev, Unreal Engine 4.26) from source on Linux end-to-end — UE4 fork, the Carla server (editor C++ modules), the LibCarla Python client wheel, and content — then verifies them against a from-source server. Optionally builds the native ROS 2 interface in (ROS2=1 → --ros2, Fast-DDS/CycloneDDS/Zenoh), which is compile-time only and cannot be enabled later. Use when the user asks to "build CARLA from source", "compile CARLA ue4-dev", "set up CARLA on Ubuntu (incl. 24.04)", "build CARLA with ROS2 support", or "produce a CARLA server + client wheel". Cooking a distributable Dist/ package is a separate skill (package-carla-ue4); running the build is run-carla-server.

日本語の概要は準備中です。原文の説明を表示しています。

carla-simulator/carla-agentic-tools42026年9月12日 更新

Builds CARLA on Unreal Engine 5.8 from the ue58-dev branch using its CMake build system — configure a preset, then build the carla-unreal / carla-unreal-editor / carla-python-api-install / launch targets — with ROS 2, DLSS, RSS and cook-scope options, and the engine-side build of the CarlaUnreal UE 5.8 fork. There is no Makefile and no Util/BuildTools in this tree, so every UE4 `make` recipe is invalid here. Use when the user asks to "build CARLA UE5", "build ue58-dev", "compile CARLA with cmake", "rebuild the Python API", or hits a CMake configure or target failure.

日本語の概要は準備中です。原文の説明を表示しています。

carla-simulator/carla-agentic-tools42026年9月12日 更新

carla-simulator のスキルをすべて見る

このスキルの問題を報告する