This repository showcases an autonomous navigation system for the Crazyflie 2.0 quadrotor. Implemented are robust path-finding algorithms, trajectory generation, and collision avoidance mechanisms, enabling smooth and safe flights in complex 3D environments.
project 1.4
Using the Crazyflie 2.0 quadrotor, I designed and implemented an advanced autonomous navigation system. This system integrates robust path planning methods with adaptive trajectory generation techniques to ensure optimal movement in complex 3D environments.
The Crazyflie quadrotor is modeled as a sphere with a 0.25m radius, maneuvering through a 3D voxelized grid environment. This environment digitization uses specific voxel resolution, striking a balance between computation speed and precision.
I employed path planning algorithms such as Dijkstra and A*. These algorithms use the formula:
f(n) = g(n) + h(n)
Where:
f(n)represents the total cost of noden.g(n)denotes the cost to reach nodenfrom the start.h(n)is a heuristic that estimates the cost from nodento the goal, enhancing the algorithm's efficiency.
The trajectories obtained from the graph algorithms were polished to be suitable for real-world quadrotor flights. A polynomial trajectory was formulated as:
s(t) = a0 + a1t + a2t^2 + ... + an*t^n
To ensure smooth transitions between trajectory segments, minimum jerk trajectories were utilized:
J = ∫(t0 to tf) (d^3x(t)/dt^3)^2 dt
Where J is the jerk over the time interval [t0, tf].
The quadrotor's dynamic movements were represented by the kinematic equations:
v(t) = ds(t)/dt a(t) = dv(t)/dt
Here, v(t) and a(t) represent velocity and acceleration respectively. A PID controller:
u(t) = Kpe(t) + Ki∫e(t) dt + Kd*de(t)/dt
Where e(t) is the error, was employed to make real-time adjustments ensuring the quadrotor moves without colliding.
Key modules such as se3_control.py (quadrotor control) and graph_search.py (path planning) were incorporated. The primary module, world_traj.py, was central in combining trajectory generation with environmental feedback, guaranteeing seamless obstacle navigation for the quadrotor.
The simulation of the quad can be found in this video[https://youtu.be/UEcTmigXg7M] .
The actual testing of the quadcopter in the crazyflie hardware can be found here[https://youtu.be/E2RCtFpBg54]
This summary elucidates the mathematical and coding foundations underpinning the significant aspects of this project.