IROS 2014 Kinect Challenge Winner: SLAM approach used
Ғылым және технология
This video shows the SLAM approach used by the winning team of the IROS 2014 Kinect Navigation Contest.
SLAM approach explained here: github.com/introlab/rtabmap/w...
Press release: www.meetup.com/SV-ROS-users/pa...
Open Source Code / ROS package: wiki.ros.org/rtabmap_ros
Пікірлер: 14
just amazing. are there any exeutable software for this*
Have you used kinect v2? Also, will this map be accurate enough to move a robot very close to the walls(obstacles) with a gap of just 2-3 cms?
@matlabbe
6 жыл бұрын
It is K4W v1 (Xbox 360). The sensor doesn't see under ~40 cm, but up to 3.5 meters, it is relatively accurate around 1 to 3 cm.
Would it be possible to use the same loop closure method for a larger scan area, say, on the order of whole city blocks?
@matlabbe
8 жыл бұрын
+Alan B Yes it can (largest area tested is about 70 km kzread.info/dash/bejne/fGx_qad6acjJg5M.html)
How many data point of this map? I mean point clouds
@matlabbe
8 жыл бұрын
+余煒森 Dont remember exactly, but the video of the mapping is 68 sec at 10 Hz, I would say 680 nodes in the map. The point cloud for each node can be generated at different resolutions (like 1pt/cm^3).
@jamesaa007
8 жыл бұрын
+matlabbe What is a node?
@matlabbe
8 жыл бұрын
+jamesaa007 The map structure is a graph with nodes and links. It is a graph-based SLAM approach.
Is the bot autonomous or remote controlled?
@matlabbe
8 жыл бұрын
+Sharath Matada In that mapping phase, it was remote controlled. In the second part of the challenge, the robot used this map to navigate autonomously to some waypoints.
@sharathmatada4135
8 жыл бұрын
+matlabbe can the mapping be achieved autonomously?
@matlabbe
8 жыл бұрын
+Sharath Matada the robot can also navigate autonomously while mapping, so yes we could use an exploration approach.
@shreyaudupa9229
7 жыл бұрын
Can you explain how one would go about using an exploration approach? Is there already code available that does this?