Week 11 - Robotics and Incremental Bayes

0.0(0)
studied byStudied by 0 people
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
Card Sorting

1/19

encourage image

There's no tags or description

Looks like no tags are added yet.

Study Analytics
Name
Mastery
Learn
Test
Matching
Spaced

No study sessions yet.

20 Terms

1
New cards

What are the main components of a robot agent?

Sensors (to perceive the world) and effectors (to act on it), including manipulators and mobile platforms.

2
New cards

What are degrees of freedom (DOF) in robotics?

The number of independent movements a robot can make. At least 6 DOF are needed to position an end-effector in 3D space (3 for position + 3 for orientation).

3
New cards

What is a non-holonomic robot?

A robot with fewer control inputs than degrees of freedom, such as a car (2 controls but 3 DOF: x, y, orientation).

4
New cards

What are the two major sources of uncertainty in robotics?

  1. Perception (sensor errors, limited resolution); 2. Action (slippage, inaccurate actuators, mechanical noise).
5
New cards

Why do perception errors occur?

Sensors have limited resolution, field of view, and may generate false positives or false negatives due to noise or occlusions.

6
New cards

Why do action errors occur?

Actions are subject to slippage, inaccurate joint encoding, surface friction, and mechanical imperfections.

7
New cards

What is the goal of robot localization?

To estimate the robot’s current state (position and orientation) using motion models and sensor observations.

8
New cards

How is a motion model used in localization?

It predicts the next state based on the robot’s velocity and turning angle (e.g., linear and angular velocity).

9
New cards

How is a sensor model used in localization?

It estimates the likelihood of being in a state given current sensor readings (e.g., distance to obstacles).

10
New cards

What is particle filtering?

A method for robot localization where many samples (particles) represent possible locations. Their probabilities are updated based on sensor data.

11
New cards

Why does particle filtering improve over time?

As the robot moves and collects more sensor readings, unlikely positions are eliminated, and particles concentrate around the true location.

12
New cards

What is SLAM?

Simultaneous Localization and Mapping: the robot builds a map of the environment while simultaneously estimating its own location within it.

13
New cards

Why is SLAM difficult in symmetric environments?

Lack of unique landmarks causes ambiguity, making it hard to resolve the robot’s location.

14
New cards

What is configuration space in robotics?

The space of all possible positions and orientations of a robot, where motion planning takes place.

15
New cards

What is cell decomposition in motion planning?

Divides configuration space into simple cells that can be navigated easily.

16
New cards

What is skeletonization in motion planning?

Reduces the space to a graph of key points (e.g., Voronoi diagram or probabilistic roadmap) to simplify pathfinding.

17
New cards

How is Bayes’ Rule used in robotics?

To update beliefs about robot state or environment based on sensor data and prior probabilities.

18
New cards

What is the incremental Bayes update rule?

P(H) ← P(M | H) * P(H) / P(M), used to incorporate new measurements into belief updates.

19
New cards

What is an example of Bayesian inference in robotics?

Estimating the presence of an obstacle using noisy sensor data, factoring in false positive/negative rates.

20
New cards

What was the DARPA Grand Challenge?

A competition where autonomous vehicles had to navigate a desert course. Stanley, from Stanford, was the first to finish using lidar and probabilistic mapping.