Execution and Control

monitor_dynamic_disturbance

Monitors tracked points or regions during execution and raises alerts when the target is moved, bumped, or shifted.

Tool Introduction

Core parameters, trigger timing, and visual before/after demo references.

Short Explanation

Keep this monitor running during manipulation so the system can react if the scene changes under the robot.

InputVideo stream + tracked points or ROI

OutputPoint trajectories and disturbance alerts

Trigger TimingTriggered on demand after the required input files and configuration are prepared.

RuntimeLocal / edge monitor

BeforeVideo stream + tracked points or ROI

Prepare the scene, image, video, sensor stream, prompt, or configuration expected by the original project.

AfterPoint trajectories and disturbance alerts

Read the produced visualization, prediction, map, trajectory, mask, grasp pose, or other documented artifact.

Preset Example

A quick-run style example for the documentation page.

Inputtools/monitor-dynamic-disturbance/examples/stream.mp4

Promptroi_points: four tracked points around the target

ExpectedTracked point trajectories, a disturbance flag, and severity score.

Parameters And Output

Readable controls and the meaning of each returned artifact.

Parameter Explanation

video_streamfile

Live stream or buffered video used for point tracking.

roitext

Region of interest around the object being monitored.

pointspathtools/monitor-dynamic-disturbance/examples/roi_points.json

Initial point coordinates to track across frames.

inference_modeselectmock_no_weights_no_install

Selects the installed tracker backend or mock mode.

Output Explanation

point_tracks

Tracked 2D point coordinates over time.

disturbance_detected

Boolean event flag for target movement or shift.

severity

Scalar score describing the magnitude of the detected disturbance.

How To Use

Official resources, deployment steps, academic context, citation, and source-reported benchmark numbers.

Resources

CoTracker GitHubhttps://github.com/facebookresearch/co-tracker Demohttps://huggingface.co/spaces/facebook/cotracker Code Downloadhttps://github.com/facebookresearch/co-tracker/archive/refs/heads/main.zip Paperhttps://arxiv.org/abs/2307.07635

Deployment Notes

Install the selected point-tracking backend or keep the local mock fallback for wrapper testing.
Prepare a live stream, buffered clip, and initial ROI points under the examples folder.
Run the monitor in parallel with the robot execution loop.
Save point tracks and disturbance alerts under tools/monitor-dynamic-disturbance/runs/.

Relative Path Example

python tools/monitor-dynamic-disturbance/run.py --video tools/monitor-dynamic-disturbance/examples/stream.mp4 --points tools/monitor-dynamic-disturbance/examples/roi_points.json --output tools/monitor-dynamic-disturbance/runs/alerts.json

Expected Result Shape

{
  "tool": "monitor-dynamic-disturbance",
  "status": "ok",
  "results": [
    {
      "label": "Dynamic disturbance monitoring",
      "score": 0.87,
      "output": "Point trajectories and disturbance alerts"
    }
  ],
  "timing": {
    "runtime": "The submitted spreadsheet describes the wrapper as real-time; the benchmark numbers above come from the official CoTracker3 evaluation rather than a separate wrapper-specific timing run.",
    "device": "documented in source benchmark when available"
  },
  "artifacts": {
    "visualization": "tools/monitor-dynamic-disturbance/runs/visualization.png",
    "raw_predictions": "tools/monitor-dynamic-disturbance/runs/predictions.json"
  }
}

Paper figure

Academic Info

Paper identity and contribution summary.

TitleCoTracker-style dynamic disturbance monitor

AuthorsAdd authors

VenueSubmitted tool sheet / local wrapper

ContributionRuns as an execution-time guard that detects target drift and produces recovery signals for closed-loop control.

Citation

@misc{monitordynamicdisturbanceYEAR,
  title={CoTracker-style dynamic disturbance monitor},
  author={Author},
  year={YEAR},
  note={Submitted tool sheet / local wrapper},
  url={https://arxiv.org/abs/2307.07635}
}

Benchmark

Only compact, source-reported numbers are shown here.

Dataset	Metric	Value	Runtime	Source
TAP-Vid / RoboTAP official evaluation	delta_avg^vis	68.3 / 76.7 / 78.8 / 82.7 on Kinetics / DAVIS / RoboTAP / RGB-S	CoTracker3 online	Official CoTracker README
TAP-Vid / RoboTAP official evaluation	delta_avg^vis	67.8 / 76.9 / 78.0 / 85.0 on Kinetics / DAVIS / RoboTAP / RGB-S	CoTracker3 offline	Official CoTracker README

Artifacts

Official CoTracker3 evaluation table, ROI point example, disturbance severity output, and local deployment notes.

Demo Images

Visual references from the original tool. Click any image to inspect the original size.