Early Deliverable Navigation Vision

This page describes the navigation stack on ROS

See the attached pdf here - ROS by example- Groovy (the latest version) for tutorials and explanation of installing ROS and running rviz (section 7.3.2) (A ros visualisation package that receives data from the navigation stack and can use a model as specified by URDF). See specifically section 7.2.5 about 2D slam with the gmapping and amcl packages that utilise the turtle-bots fake_laser_scan (section 8.3.2) package or Pointcloud_to_laserscan (which use openni drivers) or depthimage_to_laserscan (depthimage is a much faster newer version) packages and a particle filter as defined in the paper by Dieter Fox (1999): Monte Carlo Localization: Efficient Position Estimation for Mobile Robots. There also exists a MRPT (mobile robot programming toolkit) version for reference.

Gmapping tutorials

Recording and playing back Kinect data

Tf might also be useful for tracking the robot movement

Tf tutorial

Features of rviz: link

Creating a urdf file for visualisation in rviz

Publishing odometry data

Publishing and subscribing to nodes

What I think we should use and implement SLAM into:

Rgbdslam Package by launching headless and with laser_scan_matcher/hectorslam and gmapping for odom.

Also the hectorexploration node of hectorslam for autonomous setup so in theory just let the robot roam about. I will try test it in Morse maybe, a package that simulates robots and sensors.

Using sonar for the pointcloud

Possible useful packages:

[Robot pose] (http://www.ros.org/wiki/robot_pose_ekf)

Example project using these packages and a laser scanner with the pioneer robot- Note the .launch files. And a nice overview of their computer setup.

Mexico project

Q&A pages on answers.ros.org:

Gmapping

Installation help:

Groovy Roslaunch tutorial

Random:

// Arm using ssc32 motor

// Using octomaps from RGBDSLAM