Lecture 7 - Cognitive Skills for a Robot Flashcards

1
Q

Software Engineering Robot Control

A
  • Real-time control: Supports event-based reactive, and distributed interactions between sensors, motors, and algorithms
  • Reuse: Architecture is generic and abstract from specific robot platforms.
  • Robustness: Ensures robust robot behavior and graceful degradation of task performance in case of failures
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

SPA

A

A classic version of a pipeline architecture

Uses a world model instead of real word parameters

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Subsumption Architecture

A

Roomba uses it. Rule-based.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

SPA Benefits and Issues

A

+ Integrates symbolic and non-symbolic techniques

  • Robot control slow
  • Not very robust
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Subsumption Architecture Benefits and issues

A

+ Fast due to focus on behavior and tight sensor-behavior coupling.
+ It is reactive, able to handle the dynamic world.
- Difficult to compose behaviors to achieve long-range goals.
- Almost impossible to optimize

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

3T or Layered architectures

A
  • 3T: Highest deliberative layer responsible for task-planning and achieving long- term goals of the robot with given constraints.
  • 2T: Middle executive or sequence layer responsible for choosing the current behaviors of the robot to achieve a task
  • 1T: Lowest behavioral control or skills layer responsible for controlling sensors and actuators
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Layered architectures Benefits and issues

A

+ Rich architecture, with different levels of abstractions and clear roles: Planning, execution, and control.

  • Complex: Many ways to instantiate 3T.
  • How many layers are needed?
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Social Interaction Architecture (SIA)

A

Support for:

  • emotions
  • high-level dialog
  • Model other agents
  • Use social cues
  • Explain behavior
  • Learn from interaction
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

How to implement SIA?

A
  • General purpose languages
  • Cognitives agents
  • T3
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Cognitive state

A

Has domain knowledge, beliefs and goals represented. The agent program makes use of the states.
- Usually a 2-layered architecture. (Cognitive symbolic and sub-symbolic)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Symbolic layer

A
  • Fast
  • Unconscious
  • Automatic
  • Everyday decisions
  • Error prone
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Sub-symbolic layer

A
  • Slow
  • Conscious
  • Effortful
  • Complex decisions
  • Reliable
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Persistence

A

Extend architecture to integrate database technology.

Usually for high-level knowledge representation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Cloud Robotics

A

One or more robots that rely on on either data or code from a network

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Cloud Robotics Benefits and challenges

A

+ Everything is in the cloud with easy access

  • Network unreliability
  • Privacy and security
  • Service integration
How well did you know this?
1
Not at all
2
3
4
5
Perfectly