Ten capabilities. One engine. Every result, defensible.
Counter-UAS programmes fail when the platform that models them is stitched together from disconnected tools. UAVSimNet is one engine, one data model, and one reviewer-signed evidence trail — covering every surface from an airframe's aero coefficients to an engagement's probability of kill.
Every airframe arrives with its pedigree attached.
A seeded library of counter-UAS interceptors and threat airframes, plus a custom builder behind the same validation gates. JSBSim XML round-trips cleanly. Verification flights catch instability at the airframe panel — not in a broken run three hours later.
- Preset library + custom builder with identical validation gates
- JSBSim XML import and export with an opinionated round-trip
- Verification Flight auto-runs a canned trajectory per airframe
- Aero-coefficient tables with CSV paste and live coefficient plots
The weather your operators will actually fly in.
International Standard Atmosphere, non-standard variants, log- and power-law wind profiles, Dryden and Von-Kármán turbulence, gust schedules, and visibility models that flow through to the seeker chain. Every environment ships with a citation and a confidence class — not a guess.
- ISA + non-standard atmospheres, humidity, visibility, solar position
- Dryden and Von-Kármán turbulence, scripted gust schedules
- Log-law / power-law wind shear with surface-roughness coupling
- Three seeded scenarios — Clear Day, Contested Weather, Urban Canyon
Terrain that breaks the link — before the live-fire window does.
Drop a DEM. Trace obstacles as GeoJSON. The deterministic raycaster propagates the consequences straight to the seeker — per-tick detection events, comms-denied transitions, and a heat-map of denial corridors before a single Monte Carlo sample is drawn.
- OpenTopography DEM picker with inline tile preview
- GeoJSON obstacles — polygons, buildings, no-fly boxes
- Deterministic raycaster with 4/3-Earth curvature correction
- Per-tick link events propagate into seeker + comms-denied triggers
Six doctrinal threat modes — each one auditable on its own.
Single waypoint, patrol, swarm (up to 25 agents), evasive, comms-denied, loitering terminal-commit. Each mode is a distinct schema with its own validation gate — there is no way to silently ship a hybrid that one half of the engine reads differently.
- Single, patrol, swarm, evasive, comms-denied, loitering terminal-commit
- Swarm formations up to 25 agents with configurable spacing
- Evasive jink schedule editor with live sparkline preview
- Comms-denied falls back on time triggers or on-detection events
Pick the guidance law. Defend the choice.
Pure Pursuit, Proportional Navigation, Augmented PN, CLOS, and an adaptive hybrid — each paired with first-class RF or EO-IR seeker parameters. Compare up to four profiles side-by-side on identical Monte Carlo seeds.
- Pure Pursuit, PN, Augmented PN, CLOS, adaptive hybrid
- RF and EO-IR seekers with FoV, tracking noise, lock-loss envelopes
- Lock-loss → reacquire patterns (last-known cone, widening spiral)
- Side-by-side compare of 2-4 profiles on identical seeds
Fuze to p(kill), with the same audit trail throughout.
Warhead, fuze, Pk curves with live monotonicity validation, and full salvo doctrine. Single, salvo, Shoot-Look-Shoot, Shoot-Shoot-Look — every option deterministic per seed, every BDA window governed.
- Warhead, fuze, Pk curve, salvo doctrine — one coherent model
- Proximity arming + trigger separation validated before save
- Pk-curve editor with live monotonicity check + inline Pk overlay
- Salvo doctrines: single, salvo, Shoot-Look-Shoot, Shoot-Shoot-Look
Every parameter, sourced. Every result, frozen.
Every value carries source, confidence class, and citation at entry. Submitting a run captures an immutable provenance snapshot, grouped by capability — surviving any subsequent library edit. What ran is what ran. Permanently.
- source · confidence · citation on every numerical parameter
- Immutable run-start snapshot persists with the run forever
- Editing a preset later does not mutate earlier runs' snapshots
- Provenance grouped by domain in the run detail + exported reports
Interactive or batch. Same engine. No drift.
The Workbench runs single scenarios interactively with diagnostic scrubbing. The batch lane sweeps Monte Carlo across thousands of seeds. Same engine drives both — bit-exact on kinematic, metric-exact on JSBSim-hybrid.
- Real-time operator workbench with step-by-step diagnostic scrubber
- Batch lane for Monte Carlo + parameter-sweep sensitivity studies
- Bit-exact kinematic replays; metric-exact JSBSim-hybrid replays
- Same engine drives both lanes — no divergence between them
A workbench teams can sign their name to.
Role-based access (Analyst / Reviewer / Admin), per-tenant workspaces, a Cmd+K command palette, and a reviewer state machine covering draft → submitted → approved / rejected / changes-requested. Navy-themed PDF + DOCX reports ship with provenance embedded.
- Role-based workbench — Analyst / Reviewer / Admin
- Reviewer state machine with explicit Request-Changes path
- Cmd+K command palette for scenario, run, library deep-links
- Navy-themed PDF + DOCX reports with embedded provenance
Find the right configuration on a workbench, not a range.
UAVSimNet evaluates a proposed interceptor configuration, quantifies its probability of kill, and explores which design changes are most likely to improve outcome probability within defined engineering constraints. The Design Optimisation Workbench ranks parameters by their influence on outcome, generates and evaluates configuration candidates against the engineering envelope, and re-runs the leaders at higher fidelity before promotion.
- Sensitivity ranking — identifies which parameters most affect the outcome and reports their direction of influence
- Candidate exploration — generates and evaluates configuration candidates against engineering constraints
- Pareto trade-offs — plots probability of kill against design cost and highlights the knee point: the configuration with the strongest improvement per unit of design change
- Confirmation runs — re-runs top candidates at higher Monte Carlo fidelity to validate recommendations before promotion
Includes confidence-interval reporting on every candidate so statistically indistinguishable configurations are flagged together.
Read on — how it works, why the evidence holds up, and what the use cases look like.
The 5-stage pipeline — assemble, validate, run, analyse, review — told at the capability level.
Validated-vs-provisional registry, citations, bit-exact replays, reviewer chain, immutable snapshots.
Three narratives — urban-canyon net capture, contested-weather salvo, group-1 swarm vs kinetic.
See it on your own data.
We brief evaluation teams on the platform, your threat profile, and the deployment shape that fits.