Probabilistic robots Flashcards
1
Q
Features of well-calibrated robot
A
- return to desired location on average
- scatter remains due to zero mean error
- error occurs incrementally
2
Q
What variable is added due to uncertainty in motion?
A
zero mean, Gaussian error term
3
Q
How do probabilistic robots work?
A
acknowledge uncertainty and use models to abstract useful info from data
4
Q
What is sensor fusion?
A
process of combining data from many sources
5
Q
What is discrete probability distribution?
A
generalises to large numbers of possible states, as bin size smaller
6
Q
What is continuous probability distribution?
A
limiting case as widths of bins in discrete histogram tend to 0
7
Q
Process of particle filtering
A
- motion prediction based on proprioceptive sensors
- measurement update based on outward looking sensors
- normalisation
- resampling
