UAV Autonomy Core

What is sensor fusion?

Combining data from multiple sensors (IMU, GPS, camera, LiDAR) to estimate the UAV’s state (position, velocity, attitude) more accurately than any single sensor.

What does an EKF do?

It estimates the UAV state by fusing sensor data with a nonlinear motion model while accounting for noise and uncertainty.

What sensors are fused in EKF (typical)?

IMU + GPS + barometer + magnetometer.

What is VIO?

Visual-Inertial Odometry—fusion of camera + IMU for position estimation, especially in GPS-denied environments.

What is MSCKF?

A vision-based Kalman filter used for VIO, tracking visual features across frames to estimate motion.

What happens when GPS degrades?

EKF uncertainty increases → system relies on VIO, optical flow, or other sensors → may trigger safety actions.

What is “state estimation”?

The process of determining position, velocity, and orientation from sensor data.

What is IMU?

Sensor combining accelerometer + gyroscope to measure motion and rotation.

Why is IMU not enough alone?

It drifts over time → needs correction from GPS or vision.

What is sensor redundancy?

Using multiple sensors of the same type to improve reliability and allow failover.

What does PX4 do?

Handles low-level flight control, EKF estimation, and control loops.

What is ArduPilot?

Open-source autopilot similar to PX4, used for UAV control and navigation.

What is Wing’s flight stack?

Hybrid system combining PX4, ArduPilot (Protopilot), and custom layers.

What are control loops?

Cascaded loops: position → velocity → attitude → motor commands.

What control method is commonly used?

PID control.

What is a companion computer?

A Linux-based onboard computer handling high-level autonomy (VIO, perception, planning).

What does ROS do?

Middleware that manages communication between modules using nodes and topics.

What is a ROS node?

A process that performs a specific function (sensor processing, control, etc.).

What is ROS used for in your system?

Sensor fusion pipelines, data synchronization, and autonomy logic.

Why is ROS useful?

Modular, scalable, easy integration of multiple sensors and systems.

What is MAVLink?

Lightweight communication protocol between UAV components.

What does MAVLink transmit?

Telemetry, commands, and state data.

What is VISION_POSITION_ESTIMATE?

MAVLink message used to send VIO position data to the flight controller.

Why is MAVLink important?

It connects the flight controller, companion computer, and ground station.

What does Wing use for communication?

MAVLink V2 + custom communication infrastructure.

What sensors are used for perception?

Cameras, LiDAR, radar, stereo vision.

What is optical flow?

Estimating motion by tracking pixel movement between frames.

What is stereo depth?

Using two cameras to estimate distance.

What is LiDAR used for?

Accurate distance measurement and mapping.

What is radar used for?

Detecting objects in poor visibility conditions.

What happens if sensors disagree?

System increases uncertainty → may trigger failover or contingency actions.

What is contingency behavior?

Emergency actions like landing, hovering, or returning home.

What is geofencing?

Virtual boundaries restricting UAV flight area.

What is redundancy-first design?

Designing systems assuming failures will happen.

What is swarm coordination?

Multiple UAVs sharing state and coordinating actions.

What is required for swarm systems?

Reliable communication + consistent state estimation across drones.

What is multi-agent coordination?

Synchronizing multiple UAVs for a shared mission.

What is your role in swarm systems?

Integrating sensor fusion pipelines using ROS and MAVLink.

What have you been working on recently?

Developing UAV autonomy systems, prototyping solutions, and pursuing contracts.

What is your strength?

Bridging real-world flight operations with autonomy engineering.

What makes your system unique?

Focus on reliability, scalability, and real-world deployment.

What challenges did you face?

Integrating systems across distributed teams and environments.

What is IEKF?

Iterative EKF for improved nonlinear estimation.

What is CKF?

Cubature Kalman Filter for nonlinear systems.

What is SRKF?

Square Root Kalman Filter for numerical stability.

Why multiple filters?

Different subsystems require different estimation methods.

PX4 vs ROS?

PX4 = control, ROS = high-level autonomy.

IMU drift solution?

Fusion with GPS or vision.

GPS denied solution?

VIO + optical flow + perception.

MAVLink role?

Communication between system components.