Positive control is not necessarily what the desired result should look like (though in most cases it is) but rather that your system produces a response when it should. So say you're trying to test a weight loss drug on some sample. You give them a known weight loss drug so that you ascertain that these people actually do lose weight when treated with a known therapy. Otherwise, your drug could work perfectly fine but because this sample just doesn't respond to any weight loss treatment, your results would suggest that your drug doesn't work.
Negative controls make sure that your system does not produce a response when it's not supposed to. So for instance, your sample could be losing weight just because they are self-conscious and know they are in a study. So if you didn't do a negative control, you would think that your drug is working when in reality, it's not.
Vehicular control controls for the effect of the method of delivery. The best example of this is the placebo effect. So say you're giving somebody the weight loss pill and the fact that they're being given a pill makes them lose weight, not the actual composition of the drug. To control for this, you feed them a sugar pill and see if they lose weight.
