search

yujinrobot / kobuki-x

Custom Kobuki Implementations
12 stars 11 forks source link

waiterbot: is the fake laserscan used for obstacle detection? #15

Open bit-pirate opened 10 years ago

bit-pirate commented 10 years ago

In my last tests it looked like it isn't. @jihoonl do you know more? Could you have a look at it when testing at CES?

jihoonl commented 10 years ago

I haven't hear about fake laser scan. It uses three observation sources. 3dsensor, bumper, and ir sensor. paremeterized here

https://github.com/yujinrobot/kobuki-x/blob/hydro-devel/waiterbot_suite/waiterbot_ctrl_nowireless/param/move_base/costmap_common_params.yaml

What is fakelaserscan?

bit-pirate commented 10 years ago

What is fakelaserscan?

With fake laser scan I mean depthimage_to_laserscan's output. I guess, that is what you name 3dsensor.

I noticed that I don't get obstacles in the costmap from the laserscan/3D sensor readings - both for the standard local planner and the elastic band planner.

I'm not sure what is causing this.

bit-pirate commented 10 years ago

I'd like to add that bumper and cliff sensor data is correctly added to the costmap (which can make the planner fail, if you lift the robot or kick it - make sure the map is cleared before the robot tries to plan a path).

bit-pirate commented 10 years ago

I just tested it again and I can say for sure that the laser (3D sensor) data is currently not used for obstacle avoidance. move_base is subscribed to the scan topic, but I couldn't yet figure out, why its not added to the costmap. @jihoonl if you find time, please take a look at this.