CS7638 Final Exam 2026 – Georgia
Tech OMSCS Artificial Intelligence for
Robotics).
What is localization for? -ANSWER- To determine a robot's location
within a certain level of accuracy
What happens to the new mean when you have two equal standard
deviation's and you combine them? -ANSWER- The new mean is
located in the middle
What happens to the new mean when you have two unequal standard
deviation's and you combine them? -ANSWER- The new mean goes
towards the smaller variance
A valid probability distribution must sum to: -ANSWER- 100%
Is the sense function calculated used Bayes Rule or Law of Total
Probability? -ANSWER- Bayes Rule (Normalized Product)
Is the move function calculated used Bayes Rule or Law of Total
Probability? -ANSWER- Law of Total Probability (Convolution)
Is measurement a product or a convolution? -ANSWER- Product
(Bayes Rule)
Is motion a product or a convolution? -ANSWER- Convolution (Law
of Total Probability)
How do you pick the Gaussian distribution with the largest sigma? -
ANSWER- A wider distribution has a larger sigma, while a skinny
distribution has the smallest sigma
Are Histogram Filters discrete or continuous? -ANSWER- Discrete
Can a Histogram Filter be multimodal, or just unimodal? -
ANSWER- Can be multimodal
When applied to robots, do you believe the Histogram Filter is exact
or approximate? -ANSWER- Approximate
, In regards to Histogram Filters, when it comes to scaling in the
number of dimensions of the state space, which of the following is the
amount of storage that must be assigned? -ANSWER- Exponential
What happens to the probabilities of movements in a cyclic world as
the number of movements approach infinity? -ANSWER- They
become equally distributed
Are Kalman Filters discrete or continuous? -ANSWER- Continuous
Can a Kalman Filter be multimodal, or just unimodal? -ANSWER-
Cannot be multimodal (Just unimodal)
In regards to Kalman Filters, when it comes to scaling in the number
of dimensions of the state space, which of the following is the amount
of storage that must be assigned? -ANSWER- Quadratic
When applied to robots, do you believe the Kalman Filter is exact or
approximate? -ANSWER- Approximate
When would a narrow Gaussian distribution (low variance) in the
state (x matrix) NOT be good for the performance of a Kalman filter?
-ANSWER- When the mean of the prediction is far away from the mean
of the measurement
How do we calculate the next position of a robot? -ANSWER- X = F *
Xn-1
The Kalman gain is the weight applied to the measurements and the
__ estimate when updating the state. -ANSWER- Predicted state /
current state
Is the Kalman gain calculated in the prediction step of the Kalman
filter? -ANSWER- False
Prediction or Measurement Step: x' = F x + u -ANSWER- Prediction
Prediction or Measurement Step: P' = F P F^T -ANSWER- Prediction
Prediction or Measurement Step: y = z - H x -ANSWER-
Measurement
Prediction or Measurement Step: S = H P H^T + R -ANSWER-
Measurement
Tech OMSCS Artificial Intelligence for
Robotics).
What is localization for? -ANSWER- To determine a robot's location
within a certain level of accuracy
What happens to the new mean when you have two equal standard
deviation's and you combine them? -ANSWER- The new mean is
located in the middle
What happens to the new mean when you have two unequal standard
deviation's and you combine them? -ANSWER- The new mean goes
towards the smaller variance
A valid probability distribution must sum to: -ANSWER- 100%
Is the sense function calculated used Bayes Rule or Law of Total
Probability? -ANSWER- Bayes Rule (Normalized Product)
Is the move function calculated used Bayes Rule or Law of Total
Probability? -ANSWER- Law of Total Probability (Convolution)
Is measurement a product or a convolution? -ANSWER- Product
(Bayes Rule)
Is motion a product or a convolution? -ANSWER- Convolution (Law
of Total Probability)
How do you pick the Gaussian distribution with the largest sigma? -
ANSWER- A wider distribution has a larger sigma, while a skinny
distribution has the smallest sigma
Are Histogram Filters discrete or continuous? -ANSWER- Discrete
Can a Histogram Filter be multimodal, or just unimodal? -
ANSWER- Can be multimodal
When applied to robots, do you believe the Histogram Filter is exact
or approximate? -ANSWER- Approximate
, In regards to Histogram Filters, when it comes to scaling in the
number of dimensions of the state space, which of the following is the
amount of storage that must be assigned? -ANSWER- Exponential
What happens to the probabilities of movements in a cyclic world as
the number of movements approach infinity? -ANSWER- They
become equally distributed
Are Kalman Filters discrete or continuous? -ANSWER- Continuous
Can a Kalman Filter be multimodal, or just unimodal? -ANSWER-
Cannot be multimodal (Just unimodal)
In regards to Kalman Filters, when it comes to scaling in the number
of dimensions of the state space, which of the following is the amount
of storage that must be assigned? -ANSWER- Quadratic
When applied to robots, do you believe the Kalman Filter is exact or
approximate? -ANSWER- Approximate
When would a narrow Gaussian distribution (low variance) in the
state (x matrix) NOT be good for the performance of a Kalman filter?
-ANSWER- When the mean of the prediction is far away from the mean
of the measurement
How do we calculate the next position of a robot? -ANSWER- X = F *
Xn-1
The Kalman gain is the weight applied to the measurements and the
__ estimate when updating the state. -ANSWER- Predicted state /
current state
Is the Kalman gain calculated in the prediction step of the Kalman
filter? -ANSWER- False
Prediction or Measurement Step: x' = F x + u -ANSWER- Prediction
Prediction or Measurement Step: P' = F P F^T -ANSWER- Prediction
Prediction or Measurement Step: y = z - H x -ANSWER-
Measurement
Prediction or Measurement Step: S = H P H^T + R -ANSWER-
Measurement
