[ros-users] SLAM pretty much requires laser?

[Ivan Dryanovski]

Alex,

There has been plenty of work on SLAM that does not require lasers or
cameras. Here is a good, comprehensive reference to get you started:

"Simultaneous localization and mapping with limited sensing"
http://www.cs.rpi.edu/~beevek/slam.html

I have implemented SLAM on a small robot before using 6 infrared SHARP
sensors and occupancy grids - I can send you more details if you're
interested.

Now, as for ROS: To my knowledge, the navigation stack in ROS is
intended for use primarily with laser-equipped robots. I have seen
some people mention sonars on the mailing list, but I don't know what
success they've had. You might be able to use some of the packages in
the stack (obstacle avoidance? path planning?), but not the gmapping
2D slam package.

Ivan


On Tue, Sep 21, 2010 at 3:42 PM, Alex Barvo <robotnv at gmail.com> wrote:
> Hi,
>
> Several people in Trossen Robotics community are just starting to use ROS.
> Our robots are usually smaller and cheaper.
> Hokuyo and SICK sensors are way out of reach for us.
>
> We love what navigation stack can do, but it looks like it pretty much
> requires the usage of a lidar for SLAM.
> Is that a correct statement to make?
>
> Thank you,
> Alex
>
> P.S. We are aware of VSLAM, but it seems computationally intensive.
> NXT-ROS project is another example of using navigation without a laser,
> but the accuracy seems pretty limited.
>
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
>