Path Plots, Analytics, and Post-Mortem Graphing

Trajectory analysis, path planning verification, and control-loop diagnostics are supported by dedicated offline playback tooling and runtime logging analytics. This document details how log data is parsed, visualized, and evaluated to assess rover navigation performance.

---

1. The log_playback.py Visualization Suite

Located at tools/logging/log_playback.py, this Python analysis script parses the tab-delimited CSV log produced by Quill during an autonomy session (logs/<timestamp>/console_output.csv) and generates animated, synchronized multi-panel plots using matplotlib.

Script Capabilities and Parsed Telemetry

The script uses regular expressions to extract structured metrics from the unstructured and semi-structured Quill log messages:

1. 3D Positional Trajectory:
   • GPS Position: Extracted from GPS Data: (<lat> lat, <lon> lon, <alt> alt).
   • Fused Rover Pose: Extracted from Rover Pose: <lat> (lat), <lon> (lon), <alt> (alt), <deg> (degrees), GNSS/VIO FUSED? = <bool>.
   • Plotted as dual 3D trajectory subplots (Axes3D) comparing raw GPS against the visual-inertial fused pose.
2. Heading and Compass Alignment:
   • Compass Data: Extracted from Incoming Compass Data: <heading>.
   • Plotted alongside the fused pose heading to detect local magnetic anomalies, declination drift, or IMU yaw lag.
3. GNSS Accuracy and Fix Quality:
   • Extracted from Incoming Accuracy Data: (2D: <val>, 3D: <val>, Compass: <val>, FIX_TYPE: <fix>).
   • Tracks 2D horizontal accuracy, 3D spatial accuracy, and fix status across the run, highlighting GPS degradation under satellite occlusion.
4. Thread Framerate Monitors:
   • Extracted from periodic Threads FPS messages.
   • Tracks the performance of 11 concurrent threads simultaneously:
     • main_process_fps
     • main_cam_fps
     • left_cam_fps
     • right_cam_fps
     • ground_cam_fps
     • main_detector_fps
     • left_detector_fps
     • right_detector_fps
     • state_machine_fps
     • rovecomm_udp_fps
     • rovecomm_tcp_fps
5. Drivetrain Power and State Overlay:
   • Extracted from Driving at: (<left_power>, <right_power>) and Current State: <state_name>.
   • Displays real-time dynamic bar graphs of left and right motor efforts (scaled between -1.0 and +1.0) correlated directly with active state machine phases (Navigating, ApproachingMarker, Stuck, etc.).
6. Waypoint Queue Transitions:
   • Extracts waypoint additions and queue resets to demarcate leg boundaries visually along the timeline.

Usage

Run the playback script from within the autonomy workspace:

python3 tools/logging/log_playback.py path/to/logs/2026-09-08_15-30-00/console_output.csv

---

2. Real-Time Path Tracking via VisualizationHandler

While log_playback.py operates offline after a run, real-time spatial trajectories and planned paths are maintained dynamically by the VisualizationHandler:

• Path History (m_vPathHistory):
  • Accumulates DisplayPoint structs containing Easting, Northing, and Altitude relative to the local session origin m_stOriginUTM.
  • Captures the traversal score and the active rover state at each recorded point.
  • Rendered as a persistent trail in the Three.js 3D web interface.
• Planned Path (m_vPlannedPath):
  • Queried at 1 Hz from GeoPlanner::GetPlannedPath().
  • Transferred over HTTP JSON endpoints (/api/planned_path) to display upcoming A* trajectory splines and search pattern geometries.

---

3. Matplot++ Library Integration

For standalone benchmarking, algorithm evaluation, and C++ plotting routines, the build environment provides pre-compiled packages for Matplot++ (located in tools/package-builders/matplotplusplus/).

Matplot++ provides a C++ syntax mirroring MATLAB plotting functions, allowing developers to:

• Export costmap heatmaps and elevation contours directly from DuckDB point queries.
• Plot A* search trees, open sets, and closed sets during path planning algorithm tuning.
• Save high-resolution vector plots (.svg or .png) for technical design reports and competition review documentation.