dubins_path_planning

Implemented car path planning with RRT, Hybrid A*, and Dubins Path algorithms. The car model is

with the dynamics

x[t+1]     = x[t]     + cos(theta[t])
y[t+1]     = y[t]     + sin(theta[t])
theta[t+1] = theta[t] + tan(phi[t]) / L

The state variables are:

• x: horizontal position
• y: vertical position
• theta: heading angle (direction of travel)

The control variable is:

• phi ∈ [-pi/5, pi/5]: steering angle (w.r.t. the direction of travel).

Usage

$ git clone https://github.com/jhan15/dubins_path_planning.git
$ cd dubins_path_planning

# demonstarte car dynamics
$ python3 car.py

# demonstrate dubins path (shortest obstacle-free)
$ python3 dubins_path.py

# pathfinding with RRT + Dubins Path (final shot)
$ python3 rrt.py

# pathfinding with Hybrid A* + Dubins Path (final shot)
$ python3 hybrid_astar.py -heu 1 -r -e # A* heuristic + reverse + extra cost

Results

Dubins Path

Pick the shortest obstacle-free dubins path

RRT + Dubins Path

Hybrid A* + Dubins Path

w/o reversing

w/ reversing (blue tree)