Robotics Localisation /SLAM for Small UAV Navigation Research

We are looking for a robotics localisation / autonomy expert to help us select and validate the right navigation architecture for a small UAV research platform.

This is not a drone-pilot, mapping or photogrammetry project. We are looking for someone with hands-on experience in robotics localisation, visual odometry, visual-inertial odometry, SLAM, sensor fusion and UAV flight-stack integration.

The initial project is a short technical sprint. You will review our current sensor stack, test data and operating assumptions, then recommend a practical localisation architecture that can be built and tested on low-cost hardware.

Relevant experience includes:

ROS2
PX4 or ArduPilot
MAVLink / MAVROS / MAVSDK
Visual odometry / visual-inertial odometry
SLAM, including ORB-SLAM3, VINS-Fusion, OpenVINS, RTAB-Map or similar
EKF / UKF / Kalman filtering
Camera and IMU calibration
Optical flow
OpenCV
Gazebo / PX4 SITL or similar simulation tools
Raspberry Pi / Jetson-class deployment
Flight-log analysis and trajectory validation

The output should be a concise technical recommendation covering:

Recommended localisation architecture
Candidate algorithms to test
Sensor and calibration requirements
Data logging requirements
Likely failure modes
Simulation and field-test plan
30-day implementation roadmap
Clear recommendation on what not to do

The goal is to produce a practical, buildable roadmap for improving onboard localisation on a small UAV using visual, inertial and map-based cues.

Please apply only if you have direct experience with robotics localisation, SLAM/VIO, sensor fusion or UAV autonomy. In your response, please include examples of relevant previous work and briefly explain which approaches you would consider for this type of UAV localisation problem.