Testing & Results

Multiple tests (simulation and real world) were done to validate the effectivity of this software. The tests are shown below.

Test #1: Simulated Warehouse

Simulation Built in Rhinoceros 3D and Rendered / Animated in 3D Studio Max Camera 2, 3, and 6 Used for testing
		Overhead View of Warehouse (w/ Camera Assignments)
To test the initial accuracy of the system the map grid is overlaid on the ground plane. The image to the right shows that the grid accurately matches the walls and storage racks in the warehouse.
		Gridline Overlay on Warehouse Floor
Test Video Shows Forklift Warehouse Simulation from Three Different Cameras Simultaneously Camera 6: Top-Left Camera 3: Bottom-Left Camera 2: Middle-Right Forklift Tracking Features are Blue and Cyan Detected Features are Shown Overlaid in Yellow. Feature Centroids Shown by Red/Blue Dots Grid Overlay X-Coordinates are Indicated by Magenta Y-Coordinates are Indicated by Cyan		Click on Image to Play Video
Results Vehicle Position Found Very Accurately Map of Measured vs. Calculated Position Shown in Image Error Typically Less Than 1% Some Feature Classifier Break-Down at Far Distances

Overhead X-Y Plot of Vehicle Course. Actual Position (Red Dots). Position Determined by Camera2 (Green). Position Determined by Camera3 (Blue). Position Determined by Camera6 (Magenta)

Test #2: Miniature Tank

The second test was done with a small scale radio-controlled tank.
		Miniature Tank Vehicle
The experimental setup involved using two video cameras to tape the miniature tank driving on a grid. The vehicle location can be measured by inspection and compared with the camera position data.
		Experimental Test Setup
The initial gridline test was done with the experimental data. The image to the right shows that the computed X-Y gridlines coincide precisely with the actual course grid.
		Initial Gridline Test
Test Video Shows Tank from 2 View Simultaneously Forklift Tracking Features are Red and Green Detected Features are Shown Overlaid in Yellow. Feature Centroids Shown by Red/Blue Dots Grid Overlay X-Coordinates are Indicated by Magenta Y-Coordinates are Indicated by Cyan		Click on Image to Play Video
Results Vehicle Position Found Fairly Accurately Map of Measured vs. Calculated Position Shown in Image 1% to 2% Error Typically Slightly Higher Error from Sony CCD Camera at More Distant Locations Vehicle Location Lost Occasionally from Camcorder Video

Overhead X-Y Plot of Vehicle Course. Actual Position (Red Dots). Position Determined by CCD Camera (Blue). Position Determined by Camcorder (Magenta). Averaged Position from Both Cameras (Yellow).

Test #3: Radio Controlled Truck

The vehilce used for this test was an inexpensive Radio Shack remote controlled truck. The truck was fitted with an external battery pack and a wireless video camera. The course is an indoor tiled floor with 8" x 8" square tiles. These tiles served as the gridwork for this test.
		Radio Controlled Truck Vehicle
The initial gridline test was done with the experimental data. The coordinate grid calculated by the software does not run in 8" x 8" sections, but the general direction of the gridlines match well. Also, the clipping boundaries are well defined as well.
		Initial Gridline Test
Test Video Shows Tank from 2 Outside Views and One Onboard View Simultaneously Forklift Tracking Features are Red and Blue Detected Features are Shown Overlaid in Yellow. Feature Centroids Shown by Red/Blue Dots Grid Overlay X-Coordinates are Indicated by Magenta Y-Coordinates are Indicated by Cyan Range-from-Camera Coordinates Shown in Yellow		Click on Image to Play Video
Results Significant Classifier Breakdown with Distance or Lighting Changes Problematic Camera Model for Sony CCD Camera Data from Sony CCD Camera Stayed Within 3% Error

Overhead X-Y Plot of Vehicle Course. Actual Position (Red Dots). Position Determined by CCD Camera (Blue). Position Determined by Camcorder (Magenta). Averaged Position from Both Cameras (Yellow).

Test #4: Warehouse Test

The final test that was run had to be discarded because of poor lighting conditions. The vehicle's tracking features became washed out in reddish glow of the overhead lights.