Object Finder Challenge

Intermediate RoboticsIntro Python plus classes, NumPy arrays, sensor data, physical units, and supervised robot use.

Before you start

  • Complete Camera Module
  • Understand NumPy arrays, image coordinates, and depth in meters
  • Use an instructor-approved object detector

Build a camera project that finds classroom or lab objects such as a bottle, backpack, or cone and reports where each object is located.

Project Outcome

Print an object label, confidence, horizontal direction, and estimated distance. The robot remains stationary.

Camera Input

  • RGB image for object detection
  • Depth map for distance measurement
  • Image width for left, center, and right regions

Build Stages

  1. Capture and inspect one CameraFrame.
  2. Run an approved detector on frame.rgb.
  3. Find the center of each bounding box.
  4. Read depth near the box center and reject invalid values.
  5. Rank visible objects by distance or confidence.
  6. Print a clear one-line result for each object.

First Prototype

Detect one object class and print its label, confidence, and distance in meters.

What To Measure

  • Detection accuracy for each object
  • Depth stability at different distances
  • Effect of lighting and partial occlusion
  • Frequency of invalid depth measurements

Extensions

  • Build a live scoreboard.
  • Add left, center, and right direction labels.
  • Save a snapshot when a requested object appears.
  • Compare center-pixel depth with the median depth inside the bounding box.

Safety Boundary

This is a stationary camera project. Do not add searching motion until the sensing-only version is reliable and an instructor approves a separate motion design.