jnez71 / lqrrt Goto Github PK

Sometimes after traveling for a bit, things open up and re-planning right then would help a lot. How can we program in this intelligence? For example, escape behavior will switch to car behavior when a clear line of sight is found, but when would car decide it too should drop its last plan in favor of something else?

Hi Jason Nezvadovitz:
Thankyou for you project.
I want to know how to set in rviz so that i can see the GUI like the picture you paste in "Readme"?
I copy the lqrrt.rviz into .rviz fold and there is a "global status:warn -- no tf data" at rviz panel

This (extremely rare and seemingly random) error occurs during normal running:

Preparing: drive
Rotating towards goal (duration: 0.96)
OpenCV Error: Bad flag (parameter or structure field) (Unrecognized or unsupported array type) in cvGetMat, file /home/matt/Downloads/opencv-2.4.9/modules/core/src/array.cpp, line 2482
[ERROR] [WallTime: 1477538455.157950] Exception in your execute callback: /home/matt/Downloads/opencv-2.4.9/modules/core/src/array.cpp:2482: error: (-206) Unrecognized or unsupported array type in function cvGetMat

Traceback (most recent call last):
  File "/opt/ros/indigo/lib/python2.7/dist-packages/actionlib/simple_action_server.py", line 291, in executeLoop
    self.execute_callback(goal)
  File "/home/matt/mil_ws/src/lqRRT/demos/lqrrt_ros/nodes/lqrrt_node.py", line 255, in move_cb
    clean_update = self.tree_chain()
  File "/home/matt/mil_ws/src/lqRRT/demos/lqrrt_ros/nodes/lqrrt_node.py", line 310, in tree_chain
    self.goal_bias, self.sample_space, self.guide = self.select_exploration()
  File "/home/matt/mil_ws/src/lqRRT/demos/lqrrt_ros/nodes/lqrrt_node.py", line 540, in select_exploration
    area, rect = cv2.floodFill(test_flood, np.zeros((test_flood.shape[0]+2, test_flood.shape[1]+2), np.uint8), ss_goal, 69)
error: /home/matt/Downloads/opencv-2.4.9/modules/core/src/array.cpp:2482: error: (-206) Unrecognized or unsupported array type in function cvGetMat

This can be fixed by ignoring junk ogrid information that the boat "has an obstacle on it" but should probably be fixed by figuring out what is actually breaking when that kind of junk is given.

Implement David Soto's idea for refining paths by letting the current path bias a new path and keeping the one with lower cost.

The readme should be a better getting-started guide and the wiki should be comprehensive.

Hello Jason Nezvadovitz:
You project is awesome, and can you tell me which papers or algorithm you used in this project?
LQR-RRTstar?

The core planner code does not currently do RRT*. However at this point, the pure kinodynamic RRT code is clean and tested enough to start giving the * a try again.

Just do it already.