Bluetooth Low Energy Proximity sensing less than 2 feet

Light sensors aren't limited to any distance. It makes no difference whether the photons they sense came from someone's clothes 1 foot away or the sun 93 Mmiles away.

Your problem with distance is probably that the light is not focused, so the sensor is seeing the average of a large angle. When a person is close, that angle of view will be dominated by light reflecting off their clothes. When more distant, all kinds of other ambient light is mixed in.

The solution is therefore to either focus the light so that only light arriving from a narrow angle hits the sensor, or to just block the light coming from outside the view of interest. The latter can be easily done by putting the sensor at the back of a tube that is diffuse black on the inside.

I take it that the question is whether or not the radio frequency (RF) signals between the two devices will be blocked by the components in a car, when one of the devices is outside and one is inside.

Metal and other conducting materials are efficient at blocking RF. Also placing an antenna close to metal is also likely to "de-tune" the antenna, meaning it will not be as efficient on the intended frequency. Also 2.4GHz signals are not great at penetrating and have shorter range than lower frequencies. Say 900MHz would work almost 3 times as far with the same TX and RX properties. Cars can also come with metallic paints which block RF and the wind screen's coating commonly also blocks RF.

That being said, a typical bluetooth low energy device with a TX power of 0dBm and an RX sensitivity of say -90dBm gives you a theoretical 90dBm "link budget". To carry a 2.4GHz RF signal in free space for 3m, you need a link budget of 50dBm plus at least 10dBm of so called "fade margin". (Simple calculator: http://www.qsl.net/pa2ohh/jsffield.htm) Now this leaves us with a margin of 30dBm for attenuation caused by the car chassis.

RF will radiate in all directions and bounce off of materials. The question becomes, is there a way for the RF to travel between the two devices which would cause less than 30dBm of fading/attenuation. This very well may not be a straight line, but may instead be for example by bouncing off the road or getting wave-guided through the ventilation :).

Causing 30dBm attenuation is actually not very easy as the RF tends to "leak" through any and all routes. You'd be hard pressed to find locations in a car where there isn't some route that the RF finds that causes less than 30dBm of attenuation. There are some numbers for common building materials http://www.moonblink.com/store/attenuation_at_24ghz_.cfm that give you an idea of typical losses for materials.

In summary: Yes it is possible for two devices to communicate when one is placed outside of the car and one is placed inside of the car. Just don't place the antenna straight onto metal which would detune it and block a lot of RF paths.