Method

Elbrus GPU Visual SLAM [st] [IsaacElbrusGPUSLAM]


Submitted on 15 Oct. 2021 10:41 by
Dmitry Slepichev (NVIDIA)

Running time:0.007 s
Environment:Jetson AGX

Method Description:
Elbrus is a Stereo SLAM technology.

It assumes a calibrated stereo rig
(intrinsics and extrinsic of the cameras are
known, including the baseline distance) and known
lens distortion parameters
(Brown's distortion model).

Elbrus contains two
major components.
The 2D component is based on
multi-pyramid KLT feature tracking with local
average luminance compensation.
The 3D component
uses keyframes selected using 2D track average
motion and rate of disappearance criteria. The 2D
tracking of each feature is guided by the
prediction of the track's position provided by
the results of 3D tracking.

3D tracking is done via the exponential mapping
formalism for
recovering the six absolute orientation
parameters of the camera rig using standard
elements of triangulation, resectioning and
sparse bundle adjustment (SBA).

SBA runs on a
separate thread in parallel with the rest of the
Elbrus VO calculations.

For NVIDIA Xavier Elbrus can track on 144fps.
Parameters:
-lr_tracker=lk_horizontal -use_gpu=true
Latex Bibtex:
@unpublished{ELBRUS2018,
title={Realtime Stereo Visual Odometry},
author={
Alexander Korovko and
Dmitry Robustov and
Dmitry Slepichev and
Eugene Vendrovsky and
Stanislav Volodarskiy
},
}

Detailed Results

From all test sequences (sequences 11-21), our benchmark computes translational and rotational errors for all possible subsequences of length (5,10,50,100,150,...,400) meters. Our evaluation ranks methods according to the average of those values, where errors are measured in percent (for translation) and in degrees per meter (for rotation). Details for different trajectory lengths and driving speeds can be found in the plots underneath. Furthermore, the first 5 test trajectories and error plots are shown below.

Test Set Average


This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot


Sequence 11


This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



Sequence 12


This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



Sequence 13


This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



Sequence 14


This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



Sequence 15


This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot



This figure as: png eps pdf txt gnuplot





eXTReMe Tracker