lkmuk / modular-mobile-robotics Goto Github PK

and also draw some connection with Self-driving cars

implement binary search for the breakpoint search!

use this in the waypoint/ polyline module, compare the speedup

My motivation:

• use the C1 curve and other additional mission-specific heuristics for trajectory generation, which should be very useful for a variety of motion planning problems under unknown/ partially known environments/ active exploration.
• specifically, for choosing the final "goal point" (more generally, terminal state) --- instead of just choosing the "next" waypoint.
  (which may complicate the design of the trajectory generator, especially when the "inter-distance" between waypoint is varying.)

must-have:

• arc length parametrization
• possibility of adding waypoints in an online fashion
• standard interfaces for calculating the position and unit tangent vectors
• the said interfaces shall include "is_extrapolated" flag!!!
• we focus on 2D workspace but at the same time leave room for the possible extension to 3D.

should-have

• one-time indexing of the arc-length parameters for fast computation
  (associate each node with
• possibility of "projecting" a given point in

some inspirations

https://github.com/tud-cor/trajectories/blob/master/Path.cpp

consider also RViz visualization (add-on)
http://wiki.ros.org/navigation/Tutorials/Using%20rviz%20with%20the%20Navigation%20Stack

some inspirations

https://github.com/AtsushiSakai/PythonRobotics/blob/master/PathPlanning/QuinticPolynomialsPlanner/quintic_polynomials_planner.py

https://github.com/meco-group/omg-tools

https://github.com/AtsushiSakai/PythonRobotics/blob/master/PathPlanning/FrenetOptimalTrajectory/frenet_optimal_trajectory.py

• ensure smoothness at the wrapping region!
• check if the projection algorithm is still working properly at the wrapping region!
• forbid adding new points!

• Dynamic window
• MPCC