Media.Player

High-Precision Vehicle Navigation using Kalman Filter Algorithm

Technology Benefits

FEATURES AND BENEFITS OF VISUAL-AIDED INERTIAL NAVIGATION:ΓÇó Combines vision and inertial sensing (similar to human perception) ΓÇó Kalman filter-based algorithm generates pose estimation (position and orientation) information, which enables faster and more robust trackingΓÇó High accuracy and low computational complexity in highly cluttered ΓÇÿreal-worldΓÇÖ environmentsΓÇó Higher accuracy and lower cost than radar-based systems ΓÇó Operates where GPS/odometry systems may fail ΓÇó Can be integrated in existing automotive active safety systems or unmanned aerial vehicle navigation systems

Detailed Technology Description

HIGH PRECISION VEHICLE NAVIGATION USING VISUAL-AIDED INERTIAL NAVIGATION HIGH-PRECISION VEHICLE NAVIGATION SYSTEM IS A GPS ALTERNATIVEA highly precise navigation system uses visual-aided inertial navigation measurements that feeds into a unique Kalman filter based algorithm for pose estimation (position and orientation). The pose estimation algorithm can provide a unified basis for stability control, traction control, slip detection and obstacle avoidance in ground-based applications and navigation and tracking in air-based applications. The system is a GPS alternative and can operate where GPS and odometry systems fail or are denied. It can be integrated into existing automatic active safety systems and aerospace navigation systems. KALMAN FILTER BASED ALGORITHMInexpensive inertial and image sensors feed into a Kalman filter-based algorithm and enable a low-cost inrtial navigation system that has applications as a backup navigation system or as a primary navigation system. The computational requirements are significantly less than the state-of-the-art simultaneous localization and mapping technology (SLAM) and enable computational low-cost, real-time performance. The system provides real-time vehicle position, attitude, velocity and acceleration using image and inertial sensors. FEATURES AND BENEFITS OF VISUAL-AIDED INERTIAL NAVIGATION:ΓÇó Combines vision and inertial sensing (similar to human perception) ΓÇó Kalman filter-based algorithm generates pose estimation (position and orientation) information, which enables faster and more robust trackingΓÇó High accuracy and low computational complexity in highly cluttered ΓÇÿreal-worldΓÇÖ environmentsΓÇó Higher accuracy and lower cost than radar-based systems ΓÇó Operates where GPS/odometry systems may fail ΓÇó Can be integrated in existing automotive active safety systems or unmanned aerial vehicle navigation systems

Supplementary Information

Patent Number: US20090248304A1
Application Number: US2009383371A
Inventor: Roumeliotis, Stergios I. | Mourikis, Anastasios I.
Priority Date: 28 Mar 2008
Priority Number: US20090248304A1
Application Date: 23 Mar 2009
Publication Date: 1 Oct 2009
IPC Current: G01C002116
US Class: 701500 | 701220 | 701200
Assignee Applicant: Regents of the University of Minnesota,St. Paul
Title: Vision-aided inertial navigation
Usefulness: Vision-aided inertial navigation
Summary: System for processing visual information e.g. inertial sensor data, from a camera.
Novelty: Visual information e.g. inertial sensor data, processing system for camera, has processor communicatively coupled to feature tracker and generating estimate at computational complexity that is linear with tracked features

Industry

Electronics

Sub Category

3C/Gadgets

*Abstract

A system for highly precise navigation uses visual and inertial- based measurements that feeds into a unique Kalman filter based algorithm for pose estimation (position and orientation). The pose estimation algorithm can provide a unified basis for stability control, traction control, slip detection and obstacle avoidance in ground-based applications and navigation and tracking in air-based applications. The system operates where GPS and odometer systems fail or are denied and can be integrated into existing automatic active safety systems and aerospace navigation systems.

*Principal Investigator

Name: Stergios Roumeliotis, Associate Professor

Department: Computer Science and Engineering

Country/Region

USA