Merging AI and deterministic approaches for better performance: AI-Enhanced Kalman Filtering for Robust Tracking
Date:
In autonomous driving and ADAS, precise vehicle pose estimation and tracking are crucial but remain challenging due to perception accuracy, real-time processing, and environmental variability. Current methods struggle with these issues, especially in complex scenarios like occluded or distant vehicles.
To address this, a new approach combines AI-driven sensor perception with deterministic methods, using a deep-learning-based Kalman Filter to enhance temporal consistency in monocular vehicle pose estimation and track objects with competitive real-time capability. This method can leverage multiple data modalities and improves accuracy through bidirectional filtering and a learnable motion model, boosting performance across diverse driving conditions.