Drone Control By Pose Estimation Using Media Pipe
Name : AllRight Drone
Platform : DJI Telllo
Tech Stack : Python OpenCV, DJITelloPy
Purpose : Purpose Control drone with human pose detection
- Estimate human pose
- Command by pose gesture
- No controller required
- Easy to control
Tello DJI is and programmable drone perfect tor education DJITelloPy is a Python libraray using the Tello EDU SDK
- Provides various functions to control drones
OpenCV is an open-source library that includes serveral hundreds of computer vison algorithms.
- Using Python
- For Human pose estimate Pose Estimation from real-time feed plays a crucial role in various fields such ad full-body gesture control
- Detect human pose
- In real-time video Media Pipe Pose is a framework for high-fidelity body pose tracking.
- Takes input from RGB video frams
- Infers 33 3D landmarks on the whole human
- This method outperforms other methods and achieve very good results in real-time
- Using the human nose as a reference point
- Change direction while tracking location based on the center of the screen
- When the person disappearss from the screen, it rotates based on their last position
- if multiple people come in, kepp track of the first detected person drone tracked
- Error rate calculation
- Move by a constance speed margin
- If person is lost, move based on pervious_error
- Implement angle calculation of body joints to estimation pose
- Judging a pose based on the calculated angle
- Multiple joint were cross-validated to prevent malfunctions.
임시우 | siwoo5530@gmail.com 김범기 | chromato99@gmail.com