whl-diag

Autonomous driving diagnostic tools

Quick Start

1. Installation

pip install whl-diag

2. Auto-Discovery

Generate a baseline vehicle topology configuration:

whl-diag discover --out-file config/my_vehicle.yaml

3. Launch Web Service

Start the dashboard to visualize real-time progress, statistics, and results:

whl-diag serve
# Access via http://localhost:7777

4. CLI & LLM Analysis (Optional)

For automation or AI-powered root cause analysis:

# Standard CLI run
whl-diag run -c config/my_vehicle.yaml

# AI-Powered Analysis
pip install openai
export LLM_API_KEY="your-api-key"
export LLM_BASE_URL="https://api.openai.com/v1"
export LLM_MODEL="gpt-4o"

whl-diag run -c config/my_vehicle.yaml --output llm --analyze --out-file AI_Bug_Report.md

---

🧠 Probe Architecture

Diagnostics are structured into hierarchical layers to ensure systematic troubleshooting:

Layer	Focus Areas
L0 Hardware/Phy	PCIe, NIC, USB physical linkages
L1 OS/System	Thermal throttling, CPU load, MTU configs
L2 Network/Time	PTP Sync, PPS status, Packet loss
L3 Middleware	ROS2/CyberRT latencies, Shared Memory (SHM)

DAG Dependency Management

Define Directed Acyclic Graph (DAG) dependencies in config/vehicle_topology.yaml or directly in Python classes. If a dependency fails, downstream probes are skipped to prevent "ghost" errors.

class LiDARProbe(IDiagnosticProbe):
    # This probe only runs if physical link and time sync are healthy
    depends_on = ["Network Link Probe", "PTP Sync Probe"]