RoboPEPP: Vision-Based Robot Pose and Joint Angle Estimation through Embedding Predictive Pre-Training

TL;DR

RoboPEPP introduces a self-supervised, masking-based pre-training approach that leverages the robot's physical model to improve vision-based pose and joint angle estimation, especially under occlusions.

Contribution

The paper presents RoboPEPP, a novel embedding predictive pre-training method that enhances encoder understanding of robot structure for better pose estimation.

Findings

Achieves state-of-the-art accuracy in robot pose and joint angle estimation.

Demonstrates robustness to occlusions and truncations.

Requires less execution time than existing methods.

Abstract

Vision-based pose estimation of articulated robots with unknown joint angles has applications in collaborative robotics and human-robot interaction tasks. Current frameworks use neural network encoders to extract image features and downstream layers to predict joint angles and robot pose. While images of robots inherently contain rich information about the robot's physical structures, existing methods often fail to leverage it fully; therefore, limiting performance under occlusions and truncations. To address this, we introduce RoboPEPP, a method that fuses information about the robot's physical model into the encoder using a masking-based self-supervised embedding-predictive architecture. Specifically, we mask the robot's joints and pre-train an encoder-predictor model to infer the joints' embeddings from surrounding unmasked regions, enhancing the encoder's understanding of the robot's physical model. The pre-trained encoder-predictor pair, along with joint angle and keypoint prediction networks, is then fine-tuned for pose and joint angle estimation. Random masking of input during fine-tuning and keypoint filtering during evaluation further improves robustness. Our method, evaluated on several datasets, achieves the best results in robot pose and joint angle estimation while being the least sensitive to occlusions and requiring the lowest execution time.