4. Navigation

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

1/4

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.

5 Terms

1
New cards

What is the fundamental problem of Dead-Reckoning?

  • If the agent knows its current heading direction & speed it can estimate its change of position by integration.

  • Integration leads to an accumulation of uncertainty.

=> Need for “recalibration” / re-setting of the estimate.

<ul><li><p>If the agent knows its current heading direction &amp; speed it can <span style="color: yellow">estimate its change of position</span> by <span style="color: purple">integration</span>.</p></li><li><p><span style="color: purple">Integration </span>leads to an<span style="color: yellow"> accumulation of uncertainty</span>.</p></li></ul><p></p><p>=&gt; Need for “recalibration” / re-setting of the estimate.</p><p></p>
2
New cards

What is a common test to find out if an animal uses a Map?

When the animal can generalize beyond stimulus-response chaining of previous observations, f.e. take a short-cut (A→B) it previously has not.

<p>When the animal can <span style="color: yellow">generalize </span>beyond <span style="color: yellow">stimulus-response chaining</span> of previous observations, f.e. take a short-cut (A→B) it previously has not.</p><p></p>
3
New cards

What is SLAM?

  • Simultaneous Localization and Mapping.

  • Problem of learning path integration and using this to associate landmark information with locations.

<ul><li><p><span style="color: yellow">Simultaneous Localization and Mapping</span>.</p></li><li><p>Problem of learning path integration and using this to associate landmark information with locations.</p></li></ul><p></p>
4
New cards

What is the problem of Loop Closure?

  • Problem in SLAM, where an agent has reached a spot it has seen before through a different way.

  • Needs to recognize location and correct errors in map, so that the same location has the same place in the map (“close the loop”).

<ul><li><p>Problem in <span style="color: yellow">SLAM</span>, where an agent has reached a spot it <span style="color: yellow">has seen before</span> through a <span style="color: yellow">different way</span>. </p></li><li><p>Needs to recognize location and correct errors in map, so that the same location has the same place in the map (“close the loop”).</p></li></ul><p></p>
5
New cards

How does Visual Place Navigation work in Neural Dynamics?

  • Views of surrounding are stored together with current position around particular locations (“hombases”).

  • All stored views are compared to current view (with all rotations actively generated).

  • Level of correlation across multiple views within a home base generates position estimate.

<ul><li><p><span style="color: yellow">Views of surrounding are stored</span> together with current position around particular locations (“hombases”).</p></li><li><p>All stored views are <span style="color: yellow">compared to current view</span> (with all rotations actively generated).</p></li><li><p>Level of <span style="color: yellow">correlation </span>across multiple views within a home base generates <span style="color: yellow">position estimate</span>.</p></li></ul><p></p>