Kyle-Xu001 / Planning_Project

RO47005 Planning & Decision Making

Quadrotor Planning (RO47005 Project)

Simulation environment

This project is built based on this simulation environment. The already built functionalities in this environment are:

• Informed-RRT* path planner.
• Minisnap trajectory generation (solved by closed-form solution) without collision avoidance .
• A non-linear controller for path following.
• A quadrotor simulator.

Our contributions

The newly added functionalities by us are:

• PRM* path planner.
• Collision free minisnap trajectory optimization using two methods, namely corridor bounding method and waypoint insertion method.
• Kinodynamics RRT* using polynomial as steering function.
• Velocity Obstacles (currently under construction).
• Dynamic window approach (not used due to poor performance).
• Random generated map for simulation.

Please refer to our report for detailed infomation.

Dependencies

numpy, matplotlib, rtree-linux, scipy, CVXPY, CVXOPT

Usage

K-PRM* with Minisnap and Corridor Bounding Method

python runsim_prm.py

Kinodynamics RRT*

python runsim_kinorrt.py

Velocity Obstacle (still under construction)

python Quadrotor-Simulation/runsim.py

Future work

• Trajectory optimization of kinodynamics RRT*
• Using B-spline for collision avoidance
• Finer adjustment for velocity obstacle
• Try to implement model predictive control (MPC)