Case Study

Diagnose whether a planner should receive raw images or detected lane markings.

Case context: A self-driving pipeline's planner lacks lane-marking information. The team can either feed raw camera images into the planner or add a lane-marking detector whose output goes to the planner.

Question: Which design should the team choose under the task-simplicity principle, and what problem does that choice avoid?

Sample answer: The team should add the lane-marking detector and send its output to the planner. This supplies the important missing lane-marking information while avoiding a planner that must also solve the complex task of interpreting raw camera images.

Key points:

  • Choose the lane-marking detector.
  • Provide lane-marking information to the planner.
  • Avoid requiring the planner to interpret raw images.
  • Keep individual component tasks simpler.

Rubric: The response should select the lane-marking detector, identify the missing information it supplies, and explain that direct raw-image input would make the planner overly complex to build or train.

0

1

Updated 2026-07-19

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