Case Study

Selecting a Perception Module for Robot Navigation

Case context: A warehouse robot must decide whether a wheeled cart can fit through a passage. The route controller does not yet know the passage width. The engineers can either feed live hallway video directly into the controller or place a passage-width detector before it and pass the detector's output onward.

Question: Under the task-simplicity principle, which option is better, and what difficulty does it prevent?

Sample answer: Use the passage-width detector and send its result to the route controller. That gives the controller the needed width information while avoiding the much harder job of making the controller interpret raw video frames.

Key points:

  • Prefer the passage-width detector.
  • Provide passage-width information to the route controller.
  • Do not require the route controller to parse raw video.
  • Keep each module focused on a narrower task.

Rubric: The response should choose the passage-width detector, explain that it supplies the missing width information, and note that sending raw video would force the controller to handle a significantly more complex perception problem.

0

1

Updated 2026-08-12

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI