autonomous mobile robotics

Table of Contents

State of Page: inbox This page is used as an inbox to notes regarding autonoumous mobile robotics, and is not proper entry point to a well-written Zettel cluster.

I took this course in Uni of Liv, where it is coded COMP329.

The Course focused on the softwares aspects of Probablistics Robotics, including localisation, path-finding, error-handling, obstacle avoidance and many more.

I use this page to navigate through the various notes I made around this topic. Some of them are heuristics, ideas, references, and some of them are math.

1. Agent

a computer system that:

  • is situated in some environment
  • capable of autonomous action in that environemnt
  • to meet its delegated objectives

Backlinks

machine learning

Mainly, Agent with a learning function, which generally, takes in data that encodes its experience, and modify its behaviour according to that data.(which makes data science its sister discipline)

artificial intelligence is another name of algorithm

As Agent are basically one perspective of algorithm, each algorithm can be implemented on/carried out by an agent, and artificial intelligence is ideas about agents, it suffices to say that artificial intelligence is just another name of algorithm, every algorithm is intelligent, and every program anyone have ever written is some artificial intelligence.

Intelligent Agent

(Intelligent Agent)

Agent that have following 3 properties exihibited in their behaviours:

  • reactiveness
  • proactiveness
  • social ability

(2024 > 02 - February > <2024-02-17 Sat>)

in 02/17/24 to 02/18/24, created 594 nodes (this is not accurate, I think that probably is the day I move everything from /home/Notes to /home/Dropbox/Notes)

2. Agent decides

  • an agent decide what action to perform
  • an agent decide when to perform an action

Backlinks

(2024 > 02 - February > <2024-02-17 Sat>)

in 02/17/24 to 02/18/24, created 594 nodes (this is not accurate, I think that probably is the day I move everything from /home/Notes to /home/Dropbox/Notes)

3. Autonomy

4. Autonomous Agent

5. Simple Agent

6. Agent and Objects

7. Intelligent Agent

Agent that have following 3 properties exihibited in their behaviours:

  • reactiveness
  • proactiveness
  • social ability

8. Proactiveness in Intelligent Agent

The agent take initiative to attemp the goal.

The agent is not entirely reactive to the environment.

9. Reactiveness in Intelligent Agent

The agent maintain ongoing interaction with its environment, and responds to changes in its environment.

  • reactiveness is crucial as the real-world environment can be dynamic
  • reactiveness is hard to design as information about the real-world environment can be incomplete
  • dynamic environment -> possibility of failure should be taken into account in desgin

10. Social Ability in Intelligent Agent

The agent interact with other agents in the environment in its pursuit of its goal

There are types of interactions:

  • cooperation - work together -> shared goal
  • coordination - manage interdependencies between activities
  • negotiation - agreements on common interest compromises; offer and counter-offer; take turn to use shared resources.

11. Uncertainty in robotics

  • sensor <- noise and error
  • actuator <- physical force that is not rigidly defined(being non-deterministic)

World state is therefore estimated from sensors and previous actions and states

12. Probablistic Robotics

Probablistic Robotics is a series of ideas to address robotic models with explicit representation of uncertainty utilizing calculus(rigid computation methods) of probablity theory.

There 2 big stems corresponding to 2 big stems in Robotics:

  • Perception - state estimation
  • Action - utility optimisation

13. state estimation

(sensor reading, previous state estimation, current action) -> probablistic belief of the environment(including agent itself)

14. utility optimisation

find best action(utility) to perform based on belief, towards the goal.

15. Probability Theory

15.1. Axioms of probability theory

There are 3 of them

  • \(0 \leq P(A) \leq 1\)
  • \(P( True )=1 \quad P( False )=0\)
  • \(P(A \vee B)=P(A)+P(B)-P(A \wedge B)\)

16. Footnote

  • This page is an active workspace, as I need the links while writing the notes. However them will not be same as forever

Backlinks

(2024 > 02 - February > <2024-02-17 Sat>)

in 02/17/24 to 02/18/24, created 594 nodes (this is not accurate, I think that probably is the day I move everything from /home/Notes to /home/Dropbox/Notes)

Author: Linfeng He

Created: 2024-04-05 Fri 02:23