Single Camera 3D Position Estimation

RobotAI develops AI based software that transforms any camera into a 3D measurement device. It detects objects and extracts their position and orientation from a single image, single camera.

We provide 6 numbers for any known object :

  • position X,Y,Z

  • orientation angles A,B,C

Advantages


Speed

  • No point clouds

  • Using NVIDIA GPU can run 30 frame/sec

  • Fast object tracking

Flexibility

  • Define objects for robots without coding

  • Any camera : WebCam, Industrial even Smartphone

  • Working range could be easily adapted

  • System setup

Precision

RGB camera allows to use different optics

Getting close to the object allows submillimeter position accuracy

Price

Uses low cost sensors

Can run on a computer with GPU or NVIDIA JATSON

No additional equipment

Competitive software pricing



Comparison to Active Vision Sensors

Comparing RGB and active vision sensors like RealSense or LIDARs reveals that active vision sensors have their own problems:

  • Noise

  • Shadows

  • Working distances

  • Sensitivity to materials

  • Blindness to textures

At the end - you need to find an object in the point cloud

Object Examples

Carousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel image