Operation Profiles

Ops-1: Cardboard Drone

Forge a battle-ready drone from cheap cardboard that hauls a 2kg nose-cone mounted payload over 10km at 50-60 km/h speeds. Out-engineer your rivals by dominating range, velocity, and resilience—prove cardboard can conquer the skies in a low-cost UAV showdown!

Core Mission Requirements:

Airframe: ≥70% cardboard by volume
Payload: 2kg rigid, dead mass locked in the nose-cone
Design and Performance Specifications:

#	Metric	Minimum Requirement
	Range	≥ 10 km
	Cruise Speed	50–60 km/h (Sustained ≥ 5 min)
	Navigation	GPS Waypoint Navigation or First-Person Viewpoint (FPV)
	Durability	Must survive a 2m drop test
	Wingspan	< 2.5 m
	Empty Weight	< 5 kg
	Launch Method	Hand

Ops-1 Coordinator: Dr. Himanshu Dave (himanshudave@iitj.ac.in)

Student Representative: Dev Pathak (+91 843 306 8707)

Ops-2: GPS Denied Navigation

Core Mission Requirements:

The team needs to build a quadcopter that doesn’t have a GPS signal receiving facility.
On providing the present GPS coordinates value at the start point, the quadcopter should be able to navigate to the GPS location of the destinations.
The drones need to be landed at 5 different locations apart from the start point. Participants may use other sensors like a camera, LiDAR, or IMU, or any other sensor that is not linked with satellite or communication signals.

Functionalities to be incorporated

Entering present GPS Coordinates
Entering destination GPS coordinates for 5 locations (locations will be given on the spot and will be different for each team)
After the final destination, the quadcopter should have a return to Home system.

Minimum Requirements

A quadcopter should be able to fly for 15 minutes and able to travel at least 1,000 meters.

Ops-2 Coordinator: Dr. Jayant Kumar Mohanta (jayant@iitj.ac.in)

Student Representative: Mahika Saxena(+91 895 567 1281)

Ops-3: Drone in EW Environment

Track-1: Spectrum Intelligence (RF “Shazam”)
Track-2: Direction Finding & Localization (AOA / TDOA)
Track-3: Anti-UAS RF Early Warning (Passive Detection)

Track-1: Spectrum Intelligence (RF “Shazam”)

Objective

Design and develop a Spectrum Intelligence Prototype that can detect RF signals, classify signal types, and build an evolving threat/signature library from recorded/simulated RF datasets (defensive EW / ESM).

Scope

Passive analysis of the provided IQ/spectrogram datasets
Signal detection, characterization, classification, library creation, operator outputs
No transmission/jamming and no illegal interception guidance (unclassified only).

Technical Specifications (Measured at test time)

#	Category	Technical Specifications
	Input Support (any one mandatory)	IQ: SigMF or equivalent, OR Spectrogram: .npy/.npz with metadata JSON
	Signal Detection	Detect bursts + continuous signals in time-frequency P_d ≥ 0.90 at FAR ≤ 5% on Army test dataset
	Classification	Classify into the organizer-provided label set. Macro F1 ≥ 0.80
	Threat / Signature Library	Store per-event signature (freq/BW/duration + feature vector/embedding + label + confidence) “Match” new events to library: Top-3 retrieval accuracy ≥ 0.80 on repeated emitters/waveforms
	Runtime / Compute	Offline execution (no internet) Process 10 minutes of test data in ≤ 5 minutes on a laptop-class system (CPU allowed)

Deliverables

Runnable prototype (Docker preferred)
Event log (JSON) + library artifact (SQLite/JSON)
Minimal UI (web/desktop) or CLI with clear summaries
Short technical brief (slides/report): approach + metrics + limitations

Test Method

Provide a hidden test bundle (IQ or spectrogram)
Run standardized command (fixed input/output paths)
Score on: detection P_d/FAR, classification F1, library matching accuracy, runtime

Track-2: Direction Finding & Localization (AOA / TDOA)

Objective

Design and develop a Passive RF Direction Finding (DF) Prototype that estimates bearing (Angle of Arrival) and optionally emitter location using synchronized multi-sensor datasets (defensive EW / ESM).

Scope

Works on organizer-provided multi-sensor recordings (simulation/recorded only)
Bearing estimation mandatory; localization is bonus
No transmission/jamming; unclassified only

Technical Specifications

#	Category	Technical Specifications
	Input	Minimum 2 sensors (bonus for 3+) Each sensor provides synchronized IQ or derived phase/amplitude features Sensor geometry and timing metadata provided (baseline/positions/clock offsets model)
	Bearing estimation performance	RMS bearing error ≤ 5° for SNR ≥ 10 dB RMS bearing error ≤ 10° for SNR 0–10 dB Must output confidence per estimate
	Localization (bonus)	Fuse AOA (and/or TDOA if dataset supports) CEP ≤ 200 m in organizer simulation scenario (if offered)
	Robustness	Handle intermittent bursts and multi-emitter scenes Graceful degradation + confidence reporting
	Runtime / Compute	Offline execution Process a test scenario (organizer-defined) within allotted time window (e.g., ≤ 10 minutes)

Deliverables

Runnable prototype (Docker preferred)
Output file(s): time-stamped bearing estimates + confidence; (optional) location estimates + uncertainty ellipse
Minimal visualization: bearing-over-time plot + event timeline
Technical brief: DF method (phase interferometry/beamforming/MUSIC/ML, etc.), calibration assumptions, results

Test Method

Provide hidden multi-sensor dataset + sensor geometry
Run your prototype and collect bearing/location outputs
Score: bearing RMS error, stability over time, confidence calibration, robustness to multipath-like conditions

Track-3: Anti-UAS RF Early Warning (Passive Detection)

Objective

Develop a Passive RF Early-Warning Prototype that detects and flags UAS-related RF activity (control/telemetry-like links) from provided RF datasets, generating actionable alerts (defensive monitoring only). This aligns with the broader need for RF classification, including UAS-related emissions.

Scope

Passive classification and alerting on datasets only
No countermeasure/jamming features
Unclassified and offline only.

Technical Specifications

#	Category	Technical Specifications
	Input Support (any one mandatory)	IQ (SigMF) OR spectrogram bundle
	Alerting requirements	Raise an alert within ≤ 2 seconds from signal onset (measured on test set) Binary/ternary outputs: UAS-like / Non-UAS / Unknown
	Detection performance	P_d ≥ 0.90 for UAS-like events at FAR ≤ 3%
	Classification performance	F1 ≥ 0.85 on UAS-like class Confidence score required; must support threshold tuning
	Logging & library	Store detected UAS-like signatures into a library (for re-occurrence matching) Provide event metadata: time bounds, freq/BW, confidence, summary features
	Runtime / Compute	Offline execution Must run on laptop CPU baseline; bonus for real-time throughput

Deliverables

Runnable prototype (Docker preferred)
Alert stream/log (JSON) + summary report
Minimal dashboard/CLI: alerts timeline + evidence snapshot
Technical brief: model/features, false alarm control strategy, robustness testing

Test Method

Provide hidden dataset with mixed RF clutter + UAS-like events
Run prototype using fixed command format
Score: detection P_d/FAR, alert latency, UAS-like F1, robustness to clutter/noise

Ops-3 Coordinator: Dr. Akshay Moudgil (akshaymoudgil@iitj.ac.in)

Student Representative: Manisha (+91 885 890 3414)

Common Rules

No transmission/jamming, no illegal interception guidance.
Offline, unclassified datasets only, and the prototype must be testable by the agency using a standardized run command.
Teams should clearly document assumptions, limitations, and a roadmap to increase maturity/TRL.

Prime Minister's Research Fellowship (PMRF)

Vidya Lakshmi Fulfilling Educational Aspirations

National Commission for Women (NCW) Helpline