Operations analysis

ISR drone deployment, scored under realistic mission demand.

A reproducible simulation harness that compares static, random, greedy, and task-aware patrol policies on a heterogeneous ISR fleet under priority zones and time-varying surveillance tasks.

Open policy dashboard Open policy brief Open demo brief

01 · Mission snapshot

Latest run, latest answer.

Both views auto-refresh from the most recent local run. The cards summarize the dynamic policy comparison; the rankings below show the priority-weighted trade study and the policy ranking.

Best policypriority_patrolmission-fit 0.680

Task completion1.000best policy

Mean response (steps)2.15lower is better

Weighted coverage0.562best policy

Completion lift vs static0.250best policy minus static baseline

Latest demo run

Priority-weighted trade study (static vs patrol).

demo_priority_trade_study_20260504_182930

Latest policy run

Heterogeneous fleet, dynamic tasks, four policies.

policy_comparison_dynamic_heterogeneous_20260504_182940

02 · Rankings

Two views, two questions.

The priority-weighted demo asks how to spend a homogeneous fleet against priority zones. The policy comparison asks which policy serves a heterogeneous fleet best when surveillance tasks arrive on a clock.

Priority-weighted trade study

Static and patrol over the priority demo grid.

Strategy	Drones	Radius	Mission Fit	Weighted Cov	Priority Persist
patrol	3	5	0.816	0.884	0.863
patrol	6	7	0.802	0.993	0.893
patrol	3	7	0.797	0.841	0.904
patrol	9	7	0.781	0.995	0.916
patrol	6	5	0.781	0.976	0.846

Dynamic policy comparison

Static, random patrol, greedy, priority-aware patrol on a heterogeneous fleet.

Strategy	Mission Fit	Weighted Cov	Task Service	Completion	Mean Response
priority_patrol	0.680	0.562	0.384	1.000	2.15
greedy_patrol	0.620	0.496	0.381	0.750	2.47
patrol	0.516	0.834	0.086	0.850	6.50
static	0.491	0.107	0.106	0.750	0.00

03 · Evidence

Stable showcase figures.

These plots are committed in the repo (regenerated on every run) so this page is reviewable without checking out the timestamped result folders.

Policy timeseries: coverage and task service

04 · Run it locally

One-command rebuild.

The site is fully reproducible from the local artifacts: rebuild the runs, rebuild the static figures, rebuild the page.

Commands

# 1. install
$ python -m pip install -e .

# 2. regenerate demo + policy artifacts
$ make demo
$ make policy

# 3. rebuild the live demo + serve it
$ make live-demo
$ make serve-demo

# open
http://127.0.0.1:8010/docs/live_demo/index.html

The deeper writeups live in docs/.

Role fit: docs/06_role_alignment.md
Walkthrough: docs/05_demo_walkthrough.md
Dynamic policies: docs/07_dynamic_policy_comparison.md
Results: docs/04_results_summary.md