Midquarter Update

What’s Been Done So Far
The SLAM tutorial papers have been investigated. A cheap/fast version of SLAM has been developed that calculates the robot’s current position based on previous estimates and built probabilistic map using Markov-based localization. Basically, it evaluates a bunch of random points around the previous location and score each location candidate by comparing the sensor reading and the map.

Qt4 GUI has been setup so that a OpenGL widget can be displayed. The probabilistic map has been visualized using OpenGL as a very simple mesh.

Map storage has also been implemented using Sqlite3. Different portion of the map will be loaded into memory from the database as the robot navigates. Only the active portion of the map is rendered.

Progress
A lot of core progress has been made during the past two weeks. Mapping and localization are working. Although the implementation is not graphical enough at this point, a lot more progress on graphics is expected in the future as the base functionalities of the program has been made.

The majority of the proposed time line has been met, including paper investigation, Qt4 GUI setup, software foundation implementation (Classes and Objects), SLAM algorithm. The following items are behind schedule:

The robot cannot be remote controlled by the GUI using the keyboard yet. However, this is trivial and probably not an essential part of the program and can be implemented easily.

Robot is not represented in the 3D scene yet. A configuration parser has yet to be implemented to take robot geometry mesh and parameters.

Timeline Revision
Not needed except for pushing the remote control and robot representation into week 7 milestones.

BROWSE / IN TIMELINE
« Probabilistic Mapping
» Ray Tracer Assignment: Checkpoint 5

BROWSE / IN SAME CATEGORY English
« Probabilistic Mapping
» Ray Tracer Assignment: Checkpoint 5