PoseINN: Realtime Visual-based Pose Regression and Localization with Invertible Neural Networks
This addresses the problem of computational inefficiency in pose estimation for robotics and augmented reality, though it is incremental as it builds on existing invertible neural network techniques.
The paper tackles real-time camera-based pose estimation for robotics by using invertible neural networks, achieving similar accuracy to state-of-the-art methods while being faster to train and requiring only low-resolution synthetic data.
Estimating ego-pose from cameras is an important problem in robotics with applications ranging from mobile robotics to augmented reality. While SOTA models are becoming increasingly accurate, they can still be unwieldy due to high computational costs. In this paper, we propose to solve the problem by using invertible neural networks (INN) to find the mapping between the latent space of images and poses for a given scene. Our model achieves similar performance to the SOTA while being faster to train and only requiring offline rendering of low-resolution synthetic data. By using normalizing flows, the proposed method also provides uncertainty estimation for the output. We also demonstrated the efficiency of this method by deploying the model on a mobile robot.