Create a map of the lab with the robot Pioneer and SICK LMS:2

Yesterday I made a map of the laboratory using data from the laser recorded into a bag file and running slam_gmapping node.

I found a very simple mistake in my tf_listener.cpp code and searching through the sscrovers svn I found the same one. It is approximately at line 20 and it concerns the transformation of the new point created.

The original file displays the following:

try{ geometry_msgs::PointStamped base_point;
listener.transformPoint("base_laser_link", laser_point, base_point);

The transformation is in fact supposed to be from base_laser_link frame to base_link frame. And that is why base_point is created, to store the same information as laser_point (of base_laser_link frame) but now in the desired frame that should be base_link. The result is the following:

try{ geometry_msgs::PointStamped base_point;
listener.transformPoint("base_link", laser_point, base_point);

I then ran all of the nodes in exactly the same way I had been doing it and this time a map.pgm and a map.yaml file were created.

The only difference in my procedure was that, exactly at the time I was going to play the bag file, I pushed the reset button on rviz. In spite of this, gmapping still displays the TF_OLD_DATA error. I have been reading about it and several users in the mailing list argue that they had dealt with this same problem, both when running gmapping and other nodes.

Create a map of the lab with the robot Pioneer and SICK LMS:1

1. Renato made an actualization to the RosAria node in which we can notice that the transformation of which this node is in charge is odom->base_link.

//first, we'll publish the transform over tf geometry_msgs::TransformStamped odom_trans;
odom_trans.header.stamp = current_time;
odom_trans.header.frame_id = "odom";
odom_trans.child_frame_id = "base_link"; odom_trans.transform.translation.x = x;
odom_trans.transform.translation.y = y;
odom_trans.transform.translation.z = 0.0;
odom_trans.transform.rotation = odom_quat;

//send the transform
odom_broadcaster.sendTransform(odom_trans);
//next, we'll publish the odometry message over ROS
nav_msgs::Odometry odom;
odom.header.stamp = current_time; odom.header.frame_id = "odom";
odom.child_frame_id = "base_link";

I could successfully compile it and also ran the teleoperation node I made for the pioneer.

2. During the meeting with Renato we reviewed the process of transformations that should take place when creating the map using the RosAria node and the sicklms node. We agreed that the transform tree should look like this:


3. I then ran the sicklms node and could see its readings in rviz which means that the connection was being made properly. However, when I ran rosrecord I didn't get a bag file that had a correct content. I did several attempts using rosrecord with the odom, base_laser_link and base_scan topics.

4. I saw that the topic published by sicklms node is base_scan. However, it isn't yet clear for me if this node should have a frame that participates in the transform tree. What I saw in the node's code was the following:

ros::NodeHandle nh;
ros::NodeHandle nh_ns("~");
ros::Publisher scan_pub = nh.advertise("base_scan", 1);
nh_ns.param("port", port, string("/dev/ttyUSB1"));
nh_ns.param("baud", baud, 38400);
nh_ns.param("inverted", inverted, false);
nh_ns.param("frame_id", frame_id, "laser");

I wonder if the "laser" frame should be included in the tree or it should be substituted by the base_laser_link frame.
Or perhaps the base_scan topic should be substituted by a topic of the same type.

Building a map using logged data

Based on one of the gmapping tutorials:

http://www.ros.org/wiki/slam_gmapping/Tutorials/MappingFromLoggedData#download

1. Run a core

$ roscore

2. Set use_sim_time param to true

$ rosparam set use_sim_time true

3. Run the slam_gmapping node that will take scans on the base_scan topic

$ rosrun gmapping slam_gmapping scan:=base_scan

4. Play a bag file, in this case I played the example available

$ wget http://pr.willowgarage.com/data/gmapping/basic_localization_stage.bag
$ rosbag play basic_localization_stage.bag

5. In order to watch the mapping process I ran nav_view

$ rosrun nav_view nav_view /static_map:=/dynamic_map

 6. Save map  $ rosrun map_server map_saver  [ INFO] 0.000000000: Writing map occupancy data to map.pgm [ INFO] 0.000000000: Writing map occupancy data to map.yaml [ INFO] 0.000000000: Done

This is how the publications and subscriptions of topics looks like: