🪐 Walkthrough: Simulation Pipeline Alignment (Iteration 19)

This walkthrough documents the full implementation of the Simulation Pipeline Alignment, which synchronizes the automated run.bat workflow with the high-fidelity diagnostic standards of the radar metrology suite.

🚀 1. Overview of Achievements

The simulation pipeline now behaves as a production-grade ADAS testbench, generating multi-radar data and telemetry in a single automated pass.

Multi-Radar Support: Concurrent synthesis of awrl1432 and radarbook hardware profiles.
Protocol Alignment: Topics and folder structures now match the test_shenron testbench identically.
Telemetry Integration: Automatic extraction of SNR, Noise Floor, and Detection Bins.
Seamless MCAP: Single-pass packaging of all raw point clouds, metrology heatmaps, and telemetry curves.

📡 2. Unified Hardware Namespace

Visualization topics in Foxglove are now organized by hardware profile for easy comparison:

Topic Namespace	Component	Colormap	Purpose
`/radar/native`	RAW CARLA	--	Baseline physics points.
`/radar/{type}`	Shenron PCD	--	High-fidelity rich point cloud.
`/radar/{type}/metrics`	Telemetry	Plot	SNR and Noise monitoring.
`/radar/{type}/heatmaps/ra`	RA Heatmap	Jet	Top-down spatial energy (BEV).
`/radar/{type}/heatmaps/rd`	RD Heatmap	Viridis	Range-Doppler spectrum analysis.

📊 3. Deep Telemetry & Data Structure

The system persists data in a hardware-centric hierarchy inside each session folder.

Raw Signal Dumps

Saved in data/<session>/<radar_type>/metrology/:

rd/*.npy: Log-scaled power matrices.
ra/*.npy: Polar-to-Cartesian scan-converted BEV maps.
metrics.jsonl: Standardized telemetry records for Foxglove Telemetry panels.

Performance Indicators

Real-time Streaming: PYTHONUNBUFFERED logging enables instant SSE console feedback in the Dashboard.
GPU Acceleration: Multi-radar synthesis is distributed across available CUDA cores.

🏁 4. Verification Results

Recent full-scale runs confirm Protocol Parity:

Topic Integrity: MCAP files are now 1:1 compatible with testbench dashboards.
Metric Accuracy: SNR extraction is verified against the RD spectrum across all active hardware profiles.

🛠️ 5. How to Launch

The pipeline is now optimized for "One-Click" execution via the Dashboard or CLI:

Scenario CLI

./run.bat braking --frames 200

Dashboard GUI

Simply launch dashboard.bat and select your scenario. The post-processing (Shenron + MCAP) will trigger automatically upon completion.

[!IMPORTANT] Engineering Note: The RA Heatmap has transitioned to the Jet (Radar Spectrum) colormap and includes a safety clip to prevent numerical overflow in the log-conversion stage.

Verified by Antigravity | Date: 2026-04-09

2.9 KiB Raw Blame History