What are the uncertainties one would expect if the same sensor model, say a camera were mounted on different but similar platforms? An example would be the ACC camera mounted on a number of the same vehicles. I would guess there would be small position errors,uncertainties relative to the platforms reference point or CG, but I imagine these would be small and maybe calibrated out, or removed through effective data fusion techniques, yes/no?
Added: An example to hopefully clarify what I'm asking. Say there's a a caravan of passenger cars (all the same make) and each of them has the same set of cameras installed. They all go down the road following the same path, one after the other like a train. What would be the uncertainties associated across the data collected from those cameras due to the fact they they are mounted on different vehicles? I'm thinking there might be some temporal ones, maybe some come from slight differences in their mounting locations (I'm thinking very small, but not 0), some due to the fact that there would be some error because the cars would not follow the exact path. Am I'm missing any?
Thanks...
this chapter gives some explaination regarding the model uncertaities
http://www.science.uva.nl/~arnoud/education/OOAS/fwi/Chap9.pdf
Thanks Seddik. I've seen that text before and what I'm looking for is more specific to the example above. I thought the data might be competitive, in fact I still might go down that path. Consider the scenario discussed above in a different way. If the cameras are placed on a pole facing a scene and the pole can sway. Each day the cameras are replaced with another set of the same type of cameras still looking at the same scene on the sway-able pole. There would be the slight differences from the mounting of the cameras on the pole and from the different cameras themselves (inherent tolerances), although I believe these differences would be quite small. Then there would be the differences across each collection due to the swaying (similar to different vehicles going down the same path, each one following a slightly different trajectory). There would also be temporal effects to consider. These are some of the uncertainties I'm thinking of and am wondering if I'm missing any, or if I'm even at all on the right track.
I assume that you want to combine the output of the sensors and calculate a sensor fusion. Consequently, all perceptions have to be transformed into a common coordinate system. As you already discussed, you have to consider all faulty positioning assumptions. The assembling position and orientation of the sensor generates mostly a constant displacement. It has to be merged with the localization fault of the car itself!
- Badges
- Science topic
- Similar topics
- Methodology
- Sensors