🚁 Crazyflie: Advanced Real-Time Autonomous Quadcopter Navigation

This project presents a complete autonomy pipeline for a quadrotor aerial robot, developed as part of the MEAM 6200 (Advanced Robotics) course at the University of Pennsylvania. The system integrates 3D environment mapping, graph search-based path planning, smooth trajectory generation, and a geometric nonlinear SE(3) controller to achieve stable and collision-free quadrotor flight.

The pipeline was tested both in simulation and on a Crazyflie 2.0 quadrotor in a motion capture environment. Throughout this project, we progressed from modeling the quadrotor's dynamics, to planning collision-free paths, generating feasible trajectories, and finally executing those plans in the real world.

---

🧠 System Overview

The quadrotor autonomy pipeline consists of the following key components:

1. Environment Representation
   A 3D map of the world is parsed from a JSON file and converted into a voxel-based occupancy grid, which defines free and occupied space. Margins are added to account for the quadrotor’s physical size.

2. Path Planning
   Using an A* graph search algorithm, a collision-free path is computed from the start to the goal position. The planner operates in discrete voxel space, guided by a Chebyshev heuristic.

3. Trajectory Generation
   The planned path is sparsified using an adaptive waypoint selection method and then smoothed into a continuous, dynamically feasible trajectory using minimum snap optimization.

4. Geometric Control
   A nonlinear SE(3) controller computes thrust and torques needed to track the desired trajectory. The controller converts position and orientation errors into control commands and outputs rotor speeds.

5. Simulation and Real-World Testing
   The entire stack is validated in simulation using a Crazyflie dynamics model, and later deployed in a real lab environment using a Crazyflie 2.0 quadrotor with Vicon motion capture feedback.

---

📂 Project Pipeline

1. SE(3) Control

• se3_control.py implements a geometric PD controller in SE(3).
• Tracks position and orientation using tuned gain matrices:

Kp = np.diag([6., 6., 40])
Kd = np.diag([5, 5, 10])
Kr = np.diag([200, 200, 40])
Kw = np.diag([22, 22, 10])

2. Graph Search Planning

• graph_search.py implements A* over a 26-connected grid.
• Uses Chebyshev distance as heuristic and returns a dense collision-free path.

3. Trajectory Generation

• world_traj.py:
  • Uses an adaptive waypoint selection strategy.
  • Generates minimum snap trajectories for smooth flight execution.

4. Occupancy Mapping

• occupancy_map.py builds a 3D voxel grid with inflated obstacle boundaries.
• Handles coordinate conversions between metric and voxel index space.

5. Simulation & Testing

• sandbox.py integrates the full stack:
  • Loads test map
  • Plans & generates trajectory
  • Simulates flight
  • Reports: goal reached, collisions, timing, distance

---

📊 Results

A* Path, Waypoints, and Trajectory

3D Path Visualization

Position vs. Time

Orientation vs. Time

Commands vs. Time

Flight Simulation Video

Flight Animation

Actual Flight Demonstration

Actual Flight

---

🔹 Repository Structure

📂 crazyflie-autonomous-navigation/
│── 📂 docs/               # Contains project report and results
│── 📂 code/               # All of the project source code
│── README.md              # Project overview, discussion, and results
│── LICENSE                # Open-source license (MIT recommended)
│── .gitignore             # Ignore unnecessary files (build files, logs)

📅 Final Report

See docs/report_project1.4_team6.pdf for details of the real-world testing results with Crazyflie.

---

📏 Acknowledgments

• University of Pennsylvania | MEAM 6200: Advanced Robotics
• Crazyflie Labs & Vicon Motion Capture Lab
• flightsim Python library for simulation framework

---

❌ License

This repository and its contents are shared for educational showcase purposes only. Reproduction, reuse, or redistribution of any part of this project is not permitted without explicit permission from the authors.

All rights reserved.