AUTONOMICS STEREOVISION high-quality
3D vision for service robots.
Last-mile delivery, warehousing, outdoor cleaning — robotization of these routines requires depth perception perform obstacle detection, visual odometry, SLAM.
Stereovision field trials
FPGA-based dense stereo disparity calculation along with embedded real-time subpixel-accuracy postprocessing makes AUTONOMICS STEREOVISION a perfectly suited solution for various applications.
Built-in pedestrians and vehicle detection functions provided by state-of-the-art deep learning algorithms is an essential out-of-the-box feature for robots, operating in human-robot shared environment.
Automotive-grade rugged design ensures the stability of stereo vision system calibration parameters even in severe operating conditions.
ROS compatibility enables easy integration of AUTONOMICS STEREOVISION into existing systems.
Field of view: horizontal / vertical
Stereo baseline distance
Sony Pregius IMX 265, Color
1920 x 1080 pixels
60° / 32°
Embedded stereo processing
Embedded Stereo disparity pipeline
Erroneous pixels in non-occluded areas
(with 3 pixels error threshold), %
Average disparity error
in non-occluded areas, pixels
Density: percentage of pixels
for which ground truth has been
provided by the method, %
Stereo processing max performance
Disparity map resolution, pixels
Frame rate, FPS
FPGA-based: Census, multi-window
aggregation, Semi-global Matching
with 4 paths and two passes
Uniqueness check, WTA with subpixel
accuracy (1/16), Consistency check
800х400 800x400 1200x600 1200x600
128 256 128 256
30 15 20 10
Frame rate, FPS
Traffic Light (with signal recognition)
2D bounding box, class, confidence
Ambient operating temperature
9 – 36 V DC
< 40 W
Gigabit Ethernet, CAN
75g / 5 ms
5g / 5 Hz – 500 Hz
-50 °C ... +50 °C