Dual-coding representations for robot vision programming in Tekkotsu

Abstract

We describe complementary iconic and symbolic representations for parsing the visual world. The iconic pixmap representation is operated on by an extensible set of "visual routines" (Ullman, 1984;Forbus et al., 2001). A symbolic representation, in terms of lines, ellipses, blobs, etc., is extracted from the iconic encoding, manipulated algebraically, and re-rendered iconically. The two representations are therefore duals, and iconic operations can be freely intermixed with symbolic ones. The dual-coding approach offers robot programmers a versatile collection of primitives from which to construct application-specific vision software. We describe some sample applications implemented on the Sony AIBO.

Robotics for computer scientists: what’s the big idea?

“…A powerful unifying feature in Tekkotsu is the "shape space": a collection of objects describing 2D and 3D shapes in a specific coordinate system (Touretzky, Halelamien, Tira-Thompson, Wales, & Usui, 2007). There is a separate shape space for each of the three fundamental frames of reference (camera, local, and world).…”

Seven big ideas in robotics, and how to teach them

“…• The MapBuilder [14] is responsible for vision, and for maintaining egocentric (body-centered) and allocentric (world-centered) maps of the environment.…”

“…Figure 1 shows the particle filter's state made visible using a tool called the SketchGUI [14]. The robot is visualized using the Mirage simulator.…”

The Tekkotsu robotics development environment

“…The Dual-Coding vision system [15] parses image data to extract symbolic structures such as lines and ellipses, but also to process pixel-based relationships such as edges and connected components. These representations can be converted bi-directionally, from pixels to algebraic representations and back again as needed to perform efficient operations and present final results back to the user.…”