[ros-users] [Discourse.ros.org] [TurtleBot] Fusing R200 and LIDAR data | SLAM | gmapping

[Chris Albertson]

The general solution to the problem is still unsolved.    You are not going to find a turn-key solution.

I think a key insight I got from reading is that SLAM a very different problem from collision avoidance and you need different software pipelines for each problem

It makes sense to divide obstacles into two classes "dynamic" and "static".   Anything that can be moved is "dynamic" and you treat it as if it were another vehicle.  They are all moving objects that are in your shared environment.
 
Funny that LIDAR works best for indoor static obstacles  (like walls) but outdoors LIDAR is used to track dynamic obstacles (like other cars)

You might use a single 3D depth camera for both SLAM and collision avoidance but you'd have to split the data and send it to different pipelines.   

For SLAM you would prefer the sensor to be at about human "eye level" where the indoor space is more stable and it can see over most furniture.   And as I said collision avoidance is best treated separately from SLAM





---
[Visit Topic](https://discourse.ros.org/t/fusing-r200-and-lidar-data-slam-gmapping/6653/3) or reply to this email to respond.