Comparison of Graph Fitting and Sparse Deep Learning Model for Robot Pose Estimation

The paper presents a simple, yet robust computer vision system for robot arm tracking with the use of RGB-D cameras. Tracking means to measure in real time the robot state given by three angles and with known restrictions about the robot geometry. The tracking system consists of two parts: image preprocessing and machine learning. In the machine learning part, we compare two approaches: fitting the robot pose to the point cloud and fitting the convolutional neural network model to the sparse 3D depth images. The advantage of the presented approach is direct use of the point cloud transformed to the sparse image in the network input and use of sparse convolutional and pooling layers (sparse CNN). The experiments confirm that the robot tracking is performed in real time and with an accuracy comparable to the accuracy of the depth sensor.

Automated summary

This paper presents a simple yet robust computer vision system for robot arm tracking using RGB-D cameras. It tracks the robot's state in real time, given three angles and known restrictions about the robot's geometry. The system consists of two parts: image preprocessing and machine learning. In the machine learning part, two approaches are compared: fitting the robot pose to the point cloud and fitting the convolutional neural network model to the sparse 3D depth images. The presented approach directly uses the point cloud transformed to the sparse image in the network input, utilizing sparse CNN layers. Experiments confirm real-time robot tracking with accuracy comparable to that of the depth sensor.