Robot state estimation using Synthetic RGB images

The aim of this thesis is to use a camera as a cost-effective alternative to the expensive retrofitting of sensors for state estimation of the manipulator. Determining the state of robot joints from images has always been a challenge in the field of computer vision and robotics. Thus, determining the position and velocity at which the manipulator’s joint is moving is necessary for the controller when it comes to automating using only the camera as an input source. This thesis proposes to determine the state of the manipulator using RGB images. The thesis is initially limited to evaluation using synthetically generated images, then based on the results it is tested on real camera images. The research is carried out to estimate the state of the manipulator: articulation of the joint and current angular velocity.