Color Learning on a Mobile Robot: Towards Full Autonomy under Changing Illumination

Abstract
A central goal of robotics and AI is to be able to de- ploy an agent to act autonomously in the real world over an extended period of time. It is commonly asserted that in order to do so, the agent must be able to learn to deal with unexpected environmental conditions. However an ability to learn is not suf- ficient. For true extended autonomy, an agent must also be able to recognize when to abandon its cur- rent model in favor of learning a new one; and how to learn in its current situation. This paper presents a fully implemented example of such autonomy in the context of color map learning on a vision-based mobile robot for the purpose of image segmenta- tion. Past research established the ability of a robot to learn a color map in a single fixed lighting con- dition when manually given a “curriculum,” an ac- tion sequence designed to facilitate learning. This paper introduces algorithms that enable a robot to i) devise its own curriculum; and ii) recognize when the lighting conditions have changed sufficiently to warrant learning a new color map.

Read the Paper.

Okay, super nerdy. But the investigation into artificial intelligence is fascinating. Color input and the ability to recognize inputs and make decisions about abandoning current models in favor of learning a new one... um, super cool. Autonomy at is purest, more terrifying.

Robots - Color Segmentation

According to Tekkotsu, robots can't see very well, the visual abilities are inferior to a cat, dog, and probably a rat. The company provides multiple levels of vision facilities. The levels are as follows: dealing with raw camera images, color segmented images, and color segmented connected components ("blobs").

Segmentation
The segmentation of color images simplifies the vision problem by assuming that objects are colored distinctively, and that gross color differences matter. By doing this the color and brightness variations are discarded. The variations which are missing provides many valuable cues about the shapes and textures of 3-d surfaces. An advantqage to this is that the image can be processes very rapidly, this can be important in robot applications. For the color segmentation there can be multiple color classes depending on what is to be recognized . The RGB image view shows the raw image, this is always inRGB format. The segment window shows how the robot would segment the object, derived from the color class.

A tool used to define these clases is called Easy Train. The easy train tool uses five windows, the control window allows the setup of these "classes". The color spectrum window displays every pixel from every training image in a two-dimensional color space with a hue along the horizontal axis and an intensity along the vertical.