Skip to content

Predict HEXBUG Nano micro robot moving path in a box with an obstacle

Notifications You must be signed in to change notification settings

vivianhylee/robot-path-prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Robot Path Prediction

Predict HEXBUG Nano micro robot moving path in a wooden box with an obstacle

Goal

Given a 1200 seconds moving robot video as training data, we build and train a model based on AI and machine learning method and use the model to predict moving robot trajectory in 10 different test scenario.

Introduction and Algorithm Overview

  • Each test scenario is a 58 seconds long video and recorded at 30 frames per second. We will predict the positions of the robot in each of the following 2 seconds (60 frames) after video subsequences (without any new measurement data).
  • Robot is represented by its centroid coordinates (a pair of x, y integers)
  • We use gradient boosting regression method as our model. It takes current coordinates (x, y) and velocity from specified numbers of previous time frames as input features and outputs robot coordinates (x, y) for the next frame.

Requirements

  • python 2.7
  • numpy
  • scikit learn

Results

  • Green: actual path
  • Red: prediction path

test01

test02

test03

test04

test05

test06

test07

test08

test09

About

Predict HEXBUG Nano micro robot moving path in a box with an obstacle

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages