MyroC Materials to Support a C-based Course with Scribbler 2 Robots

MyroC: Data Structure Design for Scribbler 2 Camera Images

The Scribbler 2 robot with a fluke component includes a camera, and images can be sent over Bluetooth to a paired workstation. Image size depends upon the specific fluke hardware:

• Images from the camera on the original fluke have size 192 pixels (height) by 256 pixels (width).
• The camera on the Fluke 2 can take both low- and high-resolution pictures.
  • Low-resolution images have size 266 pixels (height) by 427 pixels (width).
  • High-resolution images have size 800 pixels (height) by 1280 pixels (width).

Motivations for Pictures in MyroC

MyroC utilizes these images to meet several pedagogical goals and objectives within an introductory course on imperative problem solving and data structures.

• Since an image has height, width, and pixels, a Picture provides a natural framework for a struct, illustrating abstraction and the grouping of data into a high-level structure.
• Pixels within an image have a natural 2-dimensional structure (rows and columns), illustrating 2-dimensional arrays.
• A sequence of images translates naturally to a 1-dimensional array of pictures, illustrating the use of an array for data beyond numbers, characters, or strings.
• A picture data structure generalizes naturally to allow images of varying sizes — not just those taken by a camera.

Altogether, pictures provide a motivating context for a wide range of experiments and experiments with structs, 1-dimensional arrays, 2-dimensional arrays, and data abstraction.

A Natural Struct Definition

Pragmatically, a MyroC user may run a program on either an original fluke or a Fluke 2. Further, since a workstation may connect to multiple robots, a program may connect to one or more Scribbler 2 robots that use original flukes and one or more that use Fluke 2s. Within an introductory course, students should not have to focus on which fluke is being used for any run, so the same defined data structure (e.g., Picture) should work without change and without recompilation for either fluke. This suggests that data for an original fluke might be stored in a subset of the structure declared for a Fluke 2 image.

As an initial design, a Picture might allocate space for a high resolution image:

typedef struct {
   int height;
   int width;
   Pixel pix_array [800][1280];
} Picture;

Since a Pixel requires 3 bytes for RGB data, this Picture struct requires 3,072,008 bytes of storage. Unfortunately, the run-time stack on both Linux and Mac OX X are limited to 8,192K bytes, allowing only a Picture array of size 2 to be declared.

• On both Linux and Mac OS X, the command ulimit -s can be used to increase the space allocated to the run-time stack (e.g., ulimit -s 32768), but sizes larger than 32768 are not allowed on Mac OS X. Thus, on Mac OS X, the largest Picture array is limited to size 10.
• Global declarations provide somewhat larger array declarations, but even these are limited.
• Display of images with OpenGL requires additional memory, making the display of more than 3 images unreliable.

Altogether, use of an array of high-resolution images of this type is not helpful for an introductory course.

Alternative Structures

At least three alternative approaches can help resolve this storage problem, but each has drawbacks.

typedef struct {
   int height;
   int width;
   Pixel pix_array [266][427];
} Picture;
Picture pic1;

typedef struct {
   int height;
   int width;
   Pixel * pix_array;
} Picture;
Picture pic1;
... 
pic1.pix_array
   = malloc(sizeof(Pixel)
            *pix.height*pix.width);

int pic1_height;
int pic1_width;
  ...
int pic1_pix_array[pic1_height]
                  [pic1_width];
  ...
rDisplayPicture (pic1_height,
                 pic1_width,
                 pic1_pix_array);

Conclusion

Use of the low-resolution struct is the only option that supports all three learning goals: grouping of data, practice with 2-dimensional arrays, and example with a 1-dimensional array of images. Further, since high-resolution images are quite large, no solution can support more than 2-3 such images — greatly reducing any benefits of the alternatives for image storage.

created 21 February 2016 revised 22 February 2016
For more information, please contact Henry M. Walker at walker@cs.grinnell.edu.

The MyroC project is Copyright © 2011-2015 by Henry M. Walker. Selected materials copyright by Marge Coahran, Samuel A. Rebelsky, John David Stone, and Henry Walker and used by permission. This Web page, all laboratory exercises, and all MyroC software are licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License. Course readings are licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.