You are reading the documentation for an older Pickit release. Documentation for the latest release (2.4) can be found here.
Pickit has three general detection engines, each optimized for a different type of shape or part arrangement.
Teach is the most versatile engine, and well suited for most shapes, both simple and complex. You teach the part you want to detect by showing it to the camera or uploading a CAD model of it. It is the recommended detection engine for most applications.
Flex is meant for detecting geometric shapes in 3D (cylinders and spheres) and 2D (squares, rectangles, circles and ellipses). It can detect instances of the same shape with similar or different sizes.
Pattern is similar to Flex. It looks for 2D shapes (rectangles, squares, circles and ellipses) of known fixed size, and is especially useful for detecting parts that are aligned and touching.