It is a term used in statistics. Usually, when you measure the characteristics of a group of things, they form a characteristic pattern of dots, with a lot of them falling near the middle, and fewer and fewer as you go farther out... sometimes according to a distribution called "the normal distribution."
Sometimes, when you plot the data, there will be one or two dots that are far, far away from main mass of dots. These are called "outliers." You have to decide if they are real data or if they are some kind of mistake that should not be included. For example, suppose you are measuring the outdoor temperature every ten minutes on a cloudy day, and you record these measurements: 18.1°, 16.8°, 19.0°, 19.2°, 1.90°, 18.7°, 18.2°. The 1.90° measurement is an outlier. It seems almost impossible that it could be that cold. It seems more likely that you meant to type 19.0° and typed the decimal point in the wrong place.
If an outlier is a mistake, or bad data, or something that shouldn't have been included in the study, it's important to throw it out because it is so far away from the other points that just that one point can tip the averages a lot. On the other hand, if it's good data and you're just throwing it out because you don't like it, that's... well, it's considered cheating.
Your example is an unusual use, and I have to guess, but I think it means that most of the car designs were more or less "normal," similar to each other, variations on a theme, but that one of them was a really bold departure, far different from any of the other designs.