Robot Localisation

Question 1

What is a pose

Answer 1

Combination of position and orientation (i,j,t)

Question 2

What is the problem of robot localisation?

Answer 2

Ensuring the degrees of belief about its location give it as much information as possible

Question 3

The collections of positions form a…

Answer 3

partition

Question 4

How can we work out L_(i,j)?

Answer 4

(sum t)L_(i,j,t)

Question 5

How can we work out L_t?

Answer 5

(sum i) (sum j) L_(i,j,t)

Question 6

How do we encode the robot’s state of mind?

Answer 6

Probability distribution

Question 7

What algorithms do we need to devise for assigning the relevant degrees of belief?

Answer 7

Assigning initial degrees of belief
Updating belief based upon sensor information
Updating belief in response to its action

Question 8

How do we assign initial degrees of belief?

Answer 8

Assign all non-occupied poses an equal probability

Question 9

How do we update the degrees of belief based on sensor information?

Answer 9

p(l_(i,j,t) | o) = p(o | l_(i,j,t))p(l_(i,j,t)) / sum i’,j’,t’ p(o | l_(i’,j’,t’))p(l_(i’,j’,t’))

Question 10

What sensors does our robot have?

Answer 10

left, right, forward

Question 11

What does it mean to say a reading is noisy for us?

Answer 11

if the true reading is k, the sensor will return an integer between 0 and 2k, normally distributed with mean k and standard deviation k/3

Question 12

How do we update the degrees of belief in response to an action?

Answer 12

p(l’(i,j,t) | a_h) = sum i’,j’,t’ p(l’(i,j,t) | l_(i’,j’,t’) and a)p(l_(i’,j’t’))

we get here from:
what we want- p(l’(i,j,t) | a and l(i’,j’,t’))
p(l’(i,j,t) | a_h) = sum i’,j’,t’ p(l’(i,j,t) | l_(i’,j’,t’) and a) p(l_(i’,j’,t’) | a_h)
we know- p(l_(i’,j’,t’) | a_h) = p(l_(i’,j’,t’))