Object Tracking / Driving to an Object

Prerequisites

• Object Distance Estimation

• Autonomous Movement

Object Tracking

Our turret module provides a tutorial on object tracking as well as a codebase where object tracking is implemented (be sure to also look at the case study section in the linked module):

GitBook

Driving to an Object

Our Way

Implement the following algorithm:

• Detect the object and store its bounding box (refer to the Machine Learning Toolchain module)

• Determine the distance from the object to the camera (refer to the Object Distance Estimation module). You will only need the distance on the z and x axis for this.

• You now have the hypotenuse (distance on the z-axis) and one side length (absolute value of the distance on the x-axis) of a right triangle! Solve for the other side length through the Pythagorean theorem. This second side is your y distance!

• Add the x (horizontal) and newly calculated y distance to your robot's current position, this is the object's real-world position. Note that you have to account for the robot's orientation!

• Plug the position into your autonomous movement functions (refer to the autonomous movement sections)

Alternative Methods

This video tutorial shows you how to drive to a detected object:

The FTC samples also give an excellent tutorial on driving to a detected object:

https://github.com/FIRST-Tech-Challenge/FtcRobotController/blob/e0282fcbd3de5116132a69a1d0ed229f3ca97e0f/FtcRobotController/src/main/java/org/firstinspires/ftc/robotcontroller/external/samples/ConceptVuforiaDriveToTargetWebcam.java

Real World Example

FTC Team Hazen Robotics has implemented code that can drive to a game element:

https://github.com/HazenRobotics/freight-frenzy/blob/master/TeamCode/src/main/java/org/firstinspires/ftc/teamcode/utils/TargetPositionCalculator.java

https://github.com/HazenRobotics/freight-frenzy/blob/master/TeamCode/src/main/java/org/firstinspires/ftc/teamcode/vision/HomographyTargetDistance.java