CommonRoad Reactive-Planner
This project generates solutions to trajectory planning problems given in the CommonRoad scenario format. The trajectories are generated following the sampling-based approach in [1]. This approach plans motions by sampling a discrete set of trajectories, represented as quintic polynomials in a curvilinear (Frenét) coordinate frame. The sampled trajectories are checked for kinematic feasibility and collision with static/dynamic obstacles before selecting an optimal trajectory according to a given cost function.
[1] Werling, M., et al. "Optimal trajectory generation for dynamic street scenarios in a Frenét frame." Proc. of the IEEE Int. Conf. on Robotics and Automation, 2010, pp. 987-993.
pip install commonroad-reactive-plannerA minimal example for using the package:
from commonroad_rp.reactive_planner import ReactivePlanner
from commonroad_rp.utility.config import ReactivePlannerConfiguration
from commonroad_route_planner.route_planner import RoutePlanner
# scenario filename
filename = "ZAM_Over-1_1.xml"
# Build planner configuration object (includes loading scenario)
config = ReactivePlannerConfiguration.load(f"tutorial_config_{filename[:-4]}.yaml", filename)
config.update()
# run route planner to compute route
route_planner = RoutePlanner(config.scenario, config.planning_problem)
route = route_planner.plan_routes().retrieve_first_route()
# initialize reactive planner
planner = ReactivePlanner(config)
# set reference path and desired velocity (here we just keep the current speed)
planner.set_reference_path(route.reference_path)
planner.set_desired_velocity(current_speed=planner.x_0.velocity)
# call plan function
optimal_result = planner.plan()