Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments D. Grießbach, D. Baumbach, A. Börner, S. Zuev German Aerospace Center (DLR), Institute of Robotics and Mechatronics, Dept. of Data Processing for Optical Systems, Rutherfordstr. 2, Berlin
16
Embed
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments. D. Grießbach, D. Baumbach , A. Börner , S. Zuev German Aerospace Center (DLR), Institute of Robotics and Mechatronics, Dept. of Data Processing for Optical Systems, Rutherfordstr . 2, Berlin . - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments
D. Grießbach, D. Baumbach, A. Börner, S. Zuev
German Aerospace Center (DLR), Institute of Robotics and Mechatronics,Dept. of Data Processing for Optical Systems, Rutherfordstr. 2, Berlin
Overview
- Introduction
- Integrated Positioning System- Inertial Navigation- Stereo Vision- Real time Framework
- Experimental Results
- Conclusions and Outlook
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Introduction- Assumption:
- No single technology is able to provide accurate position and attitude - Multi-sensor approach is needed
- DLR research aims to:- Generic developments- Indoor/Outdoor capability- No infrastructure/external referencing- No maps (unknown environments)- No a priori assumptions- Passive system- Real time
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Fusion
6𝐷𝑜𝐹 →(𝑥 , 𝑦 , 𝑧 ,𝛼 , 𝛽 ,𝛾)
?
Inertial Navigation - Inertial Measurement Unit- Technology: inertial navigation, based on dead reckoning
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Experimental Results – Navigation
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Experimental Results – Navigation- Unknown indoor environment- Path of about 317 m- Covering 4 floors- Closed loop for evaluation- 21 similar runs
- Tracker uses about 60 Features- Frame to frame accuracy:
5 mm/0.2 deg (2 mm/0.1 deg for viewing axis)
- Final distance error < 1% of total path length
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Experimental Results – Closed Loop Error
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Mean 1σ Std [deg] -0.1 0.2
[deg] -0.1 0.3
[deg] 5.5 14.5
x [m] 0.4 1.0
y [m] 0.5 2.2
z [m] -0.2 1.7
absolute [m] 2.7 1.3
Experimental Results – 3D Point Cloud- Parallel processing chain with matching with Semi Global Matching (SGM)- Combined with navigation solution to generate 3D point cloud - not real time yet
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Conclusion- A framework for multi-sensor navigation was presented
- The framework was applied for visual aided inertial navigation withlow cost MEMS-components and stereo vision
- Accuracy of < 1% of the total path length (over 21 runs)- Low textured scenes cause short periods of pure inertial navigation- Uncompensated IMU errors (scale error, g-sensitivity, …)
- Additional processing of a high density 3D point cloud
Accuracy Evaluation of Stereo Vision Aided Inertial Navigation for Indoor Environments > Denis Grießbach > ISPRS 2013 > Capetown
Outlook- Creation of reference data sets (Camera, IMU, GPS, etc.) with ground truth
measurements for indoor and outdoor environments- Synchronized- Calibrated- Registered
- Advancing the modelling and calibration of low cost IMUs
- Integrating absolute position measurements- GNSS measurements for outdoor- RFID/Wi-Fi/Bluetooth measurements for indoor