CrazyPracticals (21 Spring)

Members: Yujie He, Jianhao Zheng, and Longlai Qiu

[Video] [Code] [Slide]

Goal: Autonomous Navigation and Landing for Crazyflie

In this practical, we programed based on Crazyflie 2.1 to find and precisely land on a platform with height of 10 cm by utilizing z reading from flow deck. Additionally, We also utilized sensor readings from multi-ranger deck to avoid the obstacles presented in the environment.

Pipeline

Autonomous navigation & landing	Workflow
✓ Local obstacle avoidance ✓ Grid-based coverage path planning ✓ Waypoint following ✓ A* search-based re-planning

Code

Code folder: ./code/crazyflie-lib-python/group_7/

.
├── cf_load_params.py
├── cf_search.py
├── cf_state_class.py
├── cf_utilis.py
├── overall.py
├── draw_traj_demo.py
├── logs
│   ├── overall-20210530_1930_x.csv
│   ├── overall-20210530_1930_x_half.csv
│   ├── overall-20210530_1930_y.csv
│   └── overall-20210530_1930_y_half.cs
└── readme.md

• overall.py: overall pipeline from taking off to landing.

  # -x (float) for setting initial x position
  # -y (float) for setting initial y position
  # -v (bool) for enabling visualization
  python overall.py -x 0.6 -y 0.6 -v

  

• draw_traj.py: x-y trajectory visualization with region annotation

  # --log_folder (str) for assigning input log folder
  # --logname (str) for loding log file
  # --img_folder (str) for assigning output image folder
  # -n/--name (str) for assigning output image name
  # --zone_anno (bool) for enabling region annotation
  python draw_traj_demo.py --logname overall-20210530_1930 -n cf_demo --zone_anno

  

The estimated values drift considerably after long flights. Moreover, the predicted starting position is significantly different from the starting point after the drone re-takes off.

Experimental setup

Features	Figures
✓ Size: 480 cm (W) × 120 cm (H) ✓ Starting & Landing pad - starting (x, y) = (60 cm, 60 cm) - landing pad randomly placed ✓ Circular and rectangular obstacles

Features

• Modular library for different tasks

  ├── cf_load_params.py  # parameter setting
  ├── cf_search.py       # searching functions such as, coverage planning, box edge detection, A* search
  ├── cf_state_class.py  # state estimation class for the proposed task
  └── cf_utilis.py       # utility functions, such as live plotting
  

• Utilized argparse for quick parameter adjustment and tuning

• Utilized matplotlib for real-time visualization

Acknowledgement

Thanks to Prof. Dario Floreano and TAs from LIS at EPFL for these amazing tutorials and examples!