Symbiotic Optimization of Behavior

Abstract

This project was part of the DARPA Virtual Robotics Challenge (VRC). The goal was to control a simulated Atlas robot in the Gazebo/ODE virtual environment and make it walk, drive a car, and connect a hose. Our approach was based on model - predictive control (MPC). We created a parallel simulation in our virtual environment (MuJoCo) which was able to simulate the robot dynamics much faster than real - time. This made it possible to optimize the controls online -- by exploring many candidate movement strategies and finding the one that is expected to work best starting from the currently estimated state of the robot.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 01, 2015
Accession Number
ADA617165

Entities

People

  • Emanuil Todorov

Organizations

  • University of Washington

Tags

Communities of Interest

  • Autonomy

DTIC Thesaurus Topics

  • Air Force
  • Air Force Research Laboratories
  • Algorithms
  • Computations
  • Control Systems
  • Dynamics
  • Environment
  • Government Procurement
  • Inertial Measurement Units
  • Model Predictive Control
  • Optimization
  • Physics
  • Robotics
  • Robots
  • Simulations
  • Simultaneous Localization And Mapping
  • Virtual Reality

Fields of Study

  • Computer science

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Computational Modeling and Simulation

Technology Areas

  • AI & ML
  • AI & ML - Autonomous Systems
  • AI & ML - Machine Learning Algorithms
  • Autonomy