Explain the tradeoff between feeding raw camera images directly to path-planning and adding a lane-marking detector.
Question: In a self-driving vehicle pipeline, discuss the trade-offs of feeding raw camera images directly into the path-planning module versus inserting an intermediate lane-marking detector. Specifically, address how these options impact the design principle of task simplicity and the complexity of building or training the modules.
Sample answer: Feeding raw camera images directly to path-planning violates the design principle of task simplicity. It forces the path-planning module to process a raw image input and solve a much more complex task. In contrast, adding an intermediate lane-marking detector allows the system to get missing lane information to the path-planning module without making any single component overly complex to build or train.
Key points:
- Feeding raw camera images directly to the path-planning module violates task simplicity.
- Direct routing forces the path-planning module to process raw image data and solve an excessively complex task.
- Adding an intermediate lane-marking detector routes necessary information to the path planner.
- Using the detector avoids making any single pipeline module overly complex to build or train.
Rubric: To earn full credit, the response must identify that direct raw camera image input violates task simplicity and increases path-planning complexity. It must also explain that adding a lane-marking detector preserves simple, buildable, and trainable modules.
0
1
Tags
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Data Science
Machine Learning Strategy
Related
Why is adding a lane-marking detector preferred over feeding the raw camera image directly into the path-planning module?
True or False: Feeding the raw camera image directly into the path-planning module is the recommended fix because it supplies missing lane information without adding pipeline components.
Adding an intermediate _____ component passes the missing lane information to path planning while keeping every module's task appropriately simple.
Why does Machine Learning Yearning advise against feeding the raw camera image directly into the path-planning module?
Feeding the raw camera image directly into path planning is the recommended fix when lane-marking information is missing from the pipeline.
Feeding the raw camera image into path planning violates the design principle of '_____ simplicity' described in Machine Learning Yearning.
Match each pipeline design choice or concept to its correct consequence or definition from Machine Learning Yearning.
Order the reasoning steps used in Machine Learning Yearning to fix a missing-information problem without violating task simplicity.
What is the primary benefit of adding a lane-marking detector rather than routing raw camera images to path planning?
The task simplicity principle holds that each pipeline module should handle a focused, well-scoped task rather than a highly complex one.
Adding an intermediate _____ component gives path planning the lane information it needs without making any module overly complex.
Match each module or concept to its role in the redesigned self-driving pipeline from Machine Learning Yearning.
Order the data-flow steps in the correctly redesigned self-driving pipeline that avoids violating task simplicity.
Explain the tradeoff between feeding raw camera images directly to path-planning and adding a lane-marking detector.
Redesigning a Self-Driving Pipeline to Support Missing Lane Information
Identify the primary design principle violated when feeding raw images into a path-planner.