00
In productionProintermediateEmbodied safety
The agent assumes the wrong camera resolution
Read the real camera_info instead of a hard-coded resolution.
Pairs with Embodied AI: great planner, poor perceiver in Use AI efficiently with ROS 2. Read the concept first, then diagnose it here.
Unlock this lab with ProOperator report
Detection crashes on a shape mismatch; the code assumed 1920x1080.
Ubuntu 24.04, ROS 2 Jazzy, rclpy/rclcpp, an LLM coding assistant + MCP
System boundary
Trace only the relevant path.
- 01camera_info→
- 02resolution→
- 03detectionOUT
02
How this lab works
You diagnose it. No command list.
Open the repair workspace and gather your own evidence in a real terminal. No diagnostic commands are handed to you - finding the fault is the exercise. Stuck? Progressive hints unlock inside the workspace.
Verification
Pass more than the visible symptom.
- Reads camera_info
- Detection runs
- Hidden regression behavior
- Root cause explanation