Fetch Blog

Andrew Vaziri

August 8, 2016


Dynamic Obstacle Tracking

For his internship at Fetch Robotics during the winter of 2015/2016, Andrew Vaziri worked on a new costmap layer for better robot navigation in dynamic environments. His intern video describes how the costmap layer tracks dynamic obstacles, especially other robots. Using data from the laser range finder, robots are detected using a machine learning algorithm called “Random Forests”. Each potential robot instance is then tracked using a filter which can aggregate the data across several laser range finder scans.

We have finally caught up with editing, so check back soon to see even more intern videos over the next couple of weeks.

Recent Posts

  • Elon Musk-Backed Nonprofit OpenAI Teaching Robots to Learn

    Admin

  • Programming By Demonstration

    Sarah Elliot

  • Freight500 and Freight1500 have arrived!

    Admin

  • Autonomous Mapping

    Shiloh Curtis

Read next

Elon Musk-Backed Nonprofit OpenAI Teaching Robots to Learn

OpenAI, an Elon Musk-backed nonprofit research lab based in San Francisco, recently created a new algorithm that teaches robots... Continue reading


Admin

May 22, 2017

Programming By Demonstration

One of the desires with robotic systems is to devise easier ways to setup and program the robots. One... Continue reading


Sarah Elliot

May 8, 2017