Thanks...I might be able to use the stuff in art_pilot. I'll let you know.
~Deepak
"Jack O'Quin" <
jack.oquin@gmail.com>
Sent by:
ros-users-bounces@code.ros.org
12/09/2010 02:20 PM
Please respond to
User discussions <
ros-users@code.ros.org>
To
User discussions <
ros-users@code.ros.org>
cc
ros-users-bounces@code.ros.org
Subject
Re: [ros-users] controllers for car-like robots
On Thu, Dec 9, 2010 at 2:08 PM, <
DRamachandran@hra.com> wrote:
Eitan, Tully, Jack,
Thanks for your help. I looked through the ART code, and it seems to
me they have divided up the code into task specific controllers for e.g.
following lane, u-turn etc. rather
than using a single control algorithm like DWA. It seems that most of the
Urban Challenge teams did this... is there a reason why? Is it that
generic controllers aren't good enough for Ackerman drives?
I don't know about other teams, but one of the DARPA Urban Challenge
requirements was to obey the California Drivers' Handbook when
interacting with other traffic [1]. For that, we found an hierarchical
state machine very helpful. None of it was motivated by Ackermann steering
constraints.
As we re-implement our navigation using ROS, we will probably retain the
state machine, but migrate it to smach.
Our navigation package was written in a hurry for the competition. It is
mainly reactive control for following lanes and obeying the rules of the
road. The main "planning" is a simple A* to discover a route through the
road network (in the commander node). The ROS version is not a complete
port of our 2007 code. It still lacks obstacle avoidance (we are
reimplementing that using ROS messages), but it does work well on our
car.
As mentioned before, art_nav is probably not what you want, but you're
welcome to any parts that seem useful. The Ackermann steering and servo
controls are mostly implemented in the art_pilot and art_servo classes.
Maybe some of those interfaces are worth looking at.
Our "zone" controller, written by Patrick Beeson, deals with navigating in
unstructured parking lots. It does some planning using Voronoi graphs and
other techniques. It is not yet ported to ROS, but the code is there if
you want to look at it. That may be more relevant, depending on your needs
and goals.
[1]
http://www.jopha.net/index.php/jopha/article/view/14
--
joq
_______________________________________________
ros-users mailing list
ros-users@code.ros.org
https://code.ros.org/mailman/listinfo/ros-users
(text/plain)