Robot arm pose estimation through pixel-wise part classification

We propose to frame the problem of marker-less robot arm pose estimation as a pixel-wise part classification problem. As input, we use a depth image in which each pixel is classified to be either from a particular robot part or the background. The classifier is a random decision forest trained on a large number of synthetically generated and labeled depth images. From all the training samples ending up at a leaf node, a set of offsets is learned that votes for relative joint positions. Pooling these votes over all foreground pixels and subsequent clustering gives us an estimate of the true joint positions. Due to the intrinsic parallelism of pixel-wise classification, this approach can run in super real-time and is more efficient than previous ICP-like methods. We quantitatively evaluate the accuracy of this approach on synthetic data. We also demonstrate that the method produces accurate joint estimates on real data despite being purely trained on synthetic data.

Authors

• Jeannette Bohg
• Javier Romero 0002
• Alexander Herzog
• Stefan Schaal

Keywords

• Joints
• Three-dimensional displays
• Robot sensing systems
• Estimation
• Training data
• Kinematics
• Robotic Arm
• Pose Estimation
• Random Forest
• Depth Images
• Leaf Node
• Joint Position
• Part Of The Robot
• Offset Settings
• Classification Accuracy
• Use Of Information
• F1 Score
• Image Pixels
• Visual Feedback
• Semantic Segmentation
• Joint Angles
• End-effector
• Sensor Noise
• Sparse Feature
• Camera Pose
• Human Pose Estimation
• Node Splitting
• Kinematic Structure
• Visual Servoing
• Voting Scheme
• Robot Pose
• Background Class
• Qualitative Examples
• Object Tracking