We're working on a proof of concept to get an off-the-shelf drone to move precisely using a visual target.
The goal of this project (on Upwork) is to implement an EKF filter that accurately predicts position and velocity based on previous position estimates, and possibly IMU data, and/or data available from other sensors.
- position estimate (x, y, z) every 100ms, based on image 100ms old
- IMU data from drone
- estimated position and velocity at current time
- the position data is somewhat noisy (though fairly accurate)
- position data doesn't arrive at consistent times, sometimes take 80ms, othertimes say 150ms
- how will we know if the project is a success?