Stumbling back in to deal with days of posts saying "well just look at other sensors!"
Sensor fusion is hard enough to do when you assume all your sensors are working correctly. When you add in "oh, and, toss out any sensors that you think are wrong" you end up in an unverifiable mess.
A rule of thumb is that any variable you add into the decision process introduces at least one (and usually far more than one) potential defect.
This is a poor candidate for Machine Leaning AI, because there are not enough good AND bad scenarios to train on.. so you have to fall back on algorithmic methods, and you pretty much can't short-cut that.. which means a lot of development work and validation work. TANSTAAFL..
OK, I'll shuffle back to the gate now.
...tom