Visual navigation paving the way to autonomy

 
Recorded with Realsense D455

Recorded with Realsense D455

Artisense lifts the limitations that inaccurate & unreliable or prohibitively expensive & complex positioning systems impose on your application.

  • State-of-the-art and proprietary direct Visual SLAM algorithm

  • Tight fusion of vision + IMU + GNSS for centimeter accurate positioning with and without RTK GNSS

  • Works out-of-the-box: Compatible with Realsense D455

  • Versatile and customizable towards client requirements thanks to modular hardware and software design

 
 

HOW DOES IT WORK?

Concept.png

The secret sauce:
Tight sensor fusion of Artisense’s proprietary direct visual SLAM and other optional sensor modalities.

Core of Artisense solutions is always a visual inertial sensor paired with ArtiSLAM and Artisense sensor fusion software.
We reduce R&D time for customers and if necessary build on top of this exisitng technology platform to satisfy specific requirements towards:

- hardware and power
- accuracy and robustness
- target environment
- and, of course, cost

The dev kit works out of the box and the modular software architecure allows integration of additional sensors, such as GNSS, wheel odometry, or LiDAR, always depending on the individual application and challenges.

ArtiSLAM: Several modular software components in one solution

Relative localization based on last known position
Visual Inertial Odometry maintains an understanding of the position relative to a last known location by estimating the agent’s motion in 6 degrees of freedom (6DoF) using cameras and an inertial sensor. It tracks an agent’s relative motion and optionally fuses other sensor information.

Satellite-based global localization
Satellite data and correction services inform about the absolute position of an agent on the globe down to centimeters of accuracy when a good signal is available. Global Navigation Satellite Systems (GNSS) suffer from multipath errors and are completely unavailability in tunnels or indoors.
The motion understanding from VIO improves GNSS positioning accuracy and compensate for disturbances or outages of GNSS.

Map-based relocalization
Alternatively or as a redundant addition to GNSS, relocalization in a light-weight and easy to create ArtiSLAM map layer can be used to obtain positioning information continuously in any place to remove absolute positioning errors. The deep learning-based approach works robustly even in vastly differing conditions outdoors throuhout the day and across seasons.

 
 

Customers

formatted_siemens.png
formatted_db.png
formatted_ecarx.png

Partners

formatted_septentrio.png
formatted_nvidia_inception.png

Research Partner

 
formatted_frankaemika.png
 
TUM.png