GPS-Denied UAV Localization from Video Only with Python
I am working on position estimation algorithms for GPS-denied environments; this task focuses on estimating an aircraft’s position using only visual data in situations where GPS is unavailable or unreliable.
The task constraints are quite strict:
Only camera frames are provided (no GPS, no IMU fusion by default)
The goal is to estimate the x, y, z positions in a reference coordinate system
The starting position is fixed at (0,0,0)
The camera is tilted downward (~70–90°), so this is essentially a visual odometry (VIO)-like problem without traditional sensors
For each frame, we also receive inter-frame displacement cues
The system must provide:
Estimated X, Y, Z coordinates (in meters)
A status flag (indicating whether the estimate is reliable)
There’s also a twist:
Reliable reference data is available for part of the sequence
Later, the system enters a “corrupted/faulty” phase, and the model must continue making estimates without reliable signals
The evaluation is based on:
The error between the predicted trajectory and the actual state
Individual axis errors (x, y, z)
Overall trajectory consistency
If anyone has worked on this or has knowledge of it, could you help me?