Correspondence Matrices are Underrated
This work addresses a bottleneck in point-cloud registration for applications like robotics and AR/VR, offering an incremental improvement over existing deep-learning methods.
The paper tackles the problem of point-cloud registration by hypothesizing that using correspondence error in loss functions, rather than transformation error, improves performance. The result shows that modified networks converge faster and achieve higher accuracy, especially with larger misalignments.
Point-cloud registration (PCR) is an important task in various applications such as robotic manipulation, augmented and virtual reality, SLAM, etc. PCR is an optimization problem involving minimization over two different types of interdependent variables: transformation parameters and point-to-point correspondences. Recent developments in deep-learning have produced computationally fast approaches for PCR. The loss functions that are optimized in these networks are based on the error in the transformation parameters. We hypothesize that these methods would perform significantly better if they calculated their loss function using correspondence error instead of only using error in transformation parameters. We define correspondence error as a metric based on incorrectly matched point pairs. We provide a fundamental explanation for why this is the case and test our hypothesis by modifying existing methods to use correspondence-based loss instead of transformation-based loss. These experiments show that the modified networks converge faster and register more accurately even at larger misalignment when compared to the original networks.