Lesson 106: Robot Vision | The Robot Doctor

Favorite

Support Materials

How a robot can sense the world it around it using cameras, determining the distance to objects using two cameras.

Assignment: Lesson 106

  1. If a feature on an object is located at an x-coordinate of 3000 on the left camera and at an x-coordinate of 1000 on the right camera, how far in front of the robot is the object?
  2. Why does the accuracy of this method become worse as the objects move farther away? It may help to think about what the distance is if the objects are 1 pixel from the center versus 2 pixels from the center.

Printable Handout

About

Learn how a robot can sense the world around it using cameras, including how to tell the distance to objects seen by two cameras, in this 14-minute episode. The goal of this video series is to teach the basics of Robotics: the what, why, and how—with examples—and to provide take-home problems to solve.

How do robots see? How do they transform what the cameras sense into meaningful data? In this lesson we will discuss the pinhole camera model for cameras and see how a robot can use that to find the distance to a nearby object using a stereo camera setup.

Credits: WQED, RobotWits LLC, PA Rural Robotics, Dr. Jonathan Butzke, Carnegie Mellon University

Standards
  • Light and waves (image formation as signals) (STEELS.6-8.PS.4)
  • Feature detection for navigation (STEELS.6-8.TE.6)
  • Quantitative reasoning about optics and sensing (STEELS.9-12.PS.4)
  • Relating pixel measurements to physical distances (STEELS.9-12.TE.6)
  • Ratios for scaling images (CC.2.1.8.E.1)
  • Basic statistics for image data (CC.2.4.8.B.1)
  • Trigonometry for depth calculation (CC.2.3.HS.A.9)
  • Functions for image processing algorithms (CC.2.2.HS.C.8)