Detect and Avoid (DAA) Systems

The DAA Challenge: "See and Avoid" for Unmanned Systems

In manned aviation, pilots maintain safety through "see and avoid"—visually detecting traffic and taking action to prevent collisions. This fundamental safety principle works well when humans are aboard aircraft with unobstructed forward visibility. Drones lack this capability, creating a critical gap that Detect and Avoid (DAA) systems must fill.

DAA encompasses technologies and procedures enabling drones to:

• Detect: Identify potential collision threats (aircraft, obstacles, terrain)
• Track: Monitor threat trajectory to determine collision risk
• Avoid: Maneuver to maintain safe separation or alert pilot for intervention

Effective DAA is essential for Beyond Visual Line of Sight (BVLOS) operations and safe integration with manned aviation. Without reliable DAA, drones must operate under restrictive rules limiting their utility.

DAA Performance Requirements (RTCA SC-228)

The RTCA Special Committee 228 developed Minimum Operational Performance Standards (MOPS) for DAA systems, establishing quantitative requirements:

Meeting these requirements requires sophisticated sensor fusion, signal processing, and collision prediction algorithms.

DAA Technologies and Sensor Types

1. Cooperative Detection: ADS-B and Transponders

Cooperative systems rely on other aircraft broadcasting their position, velocity, and identity:

ADS-B (Automatic Dependent Surveillance-Broadcast)

Most commercial aircraft now carry ADS-B Out transmitters broadcasting position derived from GPS. ADS-B In receivers on drones can detect these broadcasts, providing:

• Excellent range: 20+ miles depending on altitude and antenna quality
• Precise data: Position accurate to GPS levels (< 10 meters), velocity, altitude, aircraft ID
• Low cost: Receivers cost $200-2000 depending on sophistication
• Well-characterized: Mature technology with known limitations

Limitations:

• Incomplete coverage: Not all aircraft carry ADS-B (gliders, ultralights, vintage aircraft)
• No obstacle detection: Only detects aircraft with ADS-B transmitters
• Limited low-altitude coverage: Ground-based ADS-B networks may not cover low-flying aircraft

Despite limitations, ADS-B is the backbone of cooperative DAA for drones operating where manned traffic exists.

Mode C/S Transponders

Traditional SSR (Secondary Surveillance Radar) transponders respond to interrogations from ground radars or TCAS-equipped aircraft. While most airliners carry Mode S, active interrogation requires compatible equipment—passive listening doesn't work. Some advanced DAA systems include Mode C/S interrogators, but size, weight, power, and cost limit deployment on small drones.

2. Non-Cooperative Detection: Radar

Non-cooperative systems detect objects without requiring transmitters/transponders on targets:

Radar (Radio Detection and Ranging)

Radar transmits radio waves and analyzes reflections to detect objects. Different radar types suit different DAA applications:

Pulse-Doppler Radar:

• Detects moving targets by Doppler shift in reflected signal
• Good at detecting fast-moving aircraft even in clutter
• Typical range: 1-5 km for small drone-mounted units
• Cost: $5,000-50,000 depending on capability

FMCW (Frequency Modulated Continuous Wave) Radar:

• Measures range by frequency difference between transmitted and reflected signal
• Lighter, lower power than pulse radar
• Effective for collision avoidance at short-medium range (500m - 2km)
• Increasingly popular for drone DAA

Radar excels at detecting non-cooperative traffic (aircraft without transponders) and can detect terrain/obstacles. However, small drones have limited payload capacity for radar systems, and miniaturization increases cost.

3. Electro-Optical (EO) and Infrared (IR) Sensors

Camera-based systems provide visual/thermal detection:

Electro-Optical (Visual Camera)

High-resolution cameras coupled with computer vision algorithms detect aircraft visually:

• Advantages: Lightweight, low power, multi-purpose (also useful for inspection/surveillance)
• Challenges: Limited range (typically < 1km for small aircraft), weather-dependent, computationally intensive image processing

Infrared/Thermal Cameras

Detect heat signatures from aircraft engines and airframes:

• Advantages: Works at night, detects hot aircraft against cool sky background
• Challenges: Limited range, expensive cooled sensors for long-range detection

AI-Enhanced Vision

Modern DAA systems use deep learning to improve visual detection:

• Trained on thousands of aircraft images at various distances/angles/conditions
• Can distinguish aircraft from birds, clouds, and ground features
• Real-time processing using edge AI accelerators (NVIDIA Jetson, Intel Movidius)
• Achieves detection ranges of 1-3km for general aviation aircraft

4. Acoustic Detection

Microphone arrays detect aircraft engine/propeller noise:

• Useful supplement to other sensors
• Detects aircraft before visual range in some conditions
• Limited effectiveness in noisy environments or against quiet aircraft
• Primarily used for ground-based DAA systems at drone launch sites

Sensor Fusion and Integrated DAA

No single sensor provides complete DAA coverage. Effective systems combine multiple technologies through sensor fusion:

Complementary Sensor Combinations

Sensor Fusion Algorithms

Combining data from multiple sensors requires sophisticated processing:

• Track Association: Determining when different sensors detect the same object
• Kalman Filtering: Combining noisy measurements to estimate true position/velocity
• Confidence Scoring: Assessing reliability of each detection
• Threat Prioritization: Focusing pilot attention on most urgent threats

Modern DAA systems use probabilistic algorithms that inherently handle uncertainty in sensor data, providing reliable threat assessment even when individual sensors have limitations.

Well Clear Definition and Collision Avoidance Logic

Detecting traffic is only half the challenge—DAA must also determine when detected traffic poses a collision risk. The concept of "Well Clear" defines safe separation:

For Drones vs. Manned Aircraft:

• Horizontal: ≥ 2000 ft (609 m)
• Vertical: ≥ 250 ft (76 m)
• Time-based: ≥ 35 seconds to closest point of approach

DAA systems continuously compute predicted closest point of approach (CPA) for all detected traffic. When a track is projected to violate Well Clear boundaries, the system alerts the pilot or commands autonomous avoidance maneuvers.

Avoidance Maneuvers

Appropriate maneuvers depend on relative geometry:

• Vertical: Descending (for drones) is generally preferred to avoid climbing into faster traffic
• Horizontal: Right turns for converging traffic (following right-of-way rules)
• Speed: Slowing or stopping (for quadcopters) removes threat

Automated DAA systems must coordinate with UTM—avoidance maneuvers may temporarily exit approved flight volumes, requiring notification to USS.

Ground-Based DAA and UTM Integration

Not all DAA sensors must be aboard the drone. Ground-based and network-based solutions contribute to overall DAA capability:

Ground-Based Radar and Observers

Some operations use ground-based sensors providing airspace awareness:

• Radar installations scanning operation area
• Visual observers with binoculars/telescopes
• Acoustic detection arrays at fixed locations

Ground observers communicate threats to pilots via radio, enabling avoidance. While less automated than onboard DAA, ground-based approaches have enabled numerous BVLOS waivers as transitional solutions.

UTM-Provided Situational Awareness

USS can aggregate tracking data from multiple sources and provide airspace awareness to drone operators:

• ADS-B data from FAA feeds or commercial providers
• Manned aircraft flight plans and ATC coordination data
• Position reports from other UAS via Remote ID
• Temporary flight restrictions and constraints

While not meeting SC-228 standards for collision avoidance, UTM situational awareness complements onboard DAA, providing strategic awareness enabling proactive deconfliction.

Collaborative DAA via V2V Communication

Future systems may use vehicle-to-vehicle (V2V) communication enabling drones to share intent and coordinate avoidance:

• Drones broadcast planned trajectory
• Nearby drones adjust plans to maintain separation
• Distributed consensus algorithms resolve conflicts without central coordination

This approach could enable high-density operations with scalability beyond centralized UTM coordination.