Real-Time LPV-Based Non-Linear Model Predictive Control for Robust Trajectory Tracking in Autonomous Vehicles

This paper presents the development and implementation of a Model Predictive Control (MPC) framework for trajectory tracking in autonomous vehicles under diverse driving conditions. The proposed approach incorporates a modular architecture that integrates state estimation, vehicle dynamics modeling, and optimization to ensure real-time performance. The state-space equations are formulated in a Linear Parameter Varying (LPV) form, and a curvature-based tuning method is introduced to optimize weight matrices for varying trajectories. The MPC framework is implemented using the Robot Operating System (ROS) for parallel execution of state estimation and control optimization, ensuring scalability and minimal latency. Extensive simulations and real-time experiments were conducted on multiple predefined trajectories, demonstrating high accuracy with minimal cross-track and orientation errors, even under aggressive maneuvers and high-speed conditions. The results highlight the robustness and adaptability of the proposed system, achieving seamless alignment between simulated and real-world performance. This work lays the foundation for dynamic weight tuning and integration into cooperative autonomous navigation systems, paving the way for enhanced safety and efficiency in autonomous driving applications.

Lost in Context: The Influence of Context on Feature Attribution Methods for Object Recognition

Contextual information plays a critical role in object recognition models within computer vision, where changes in context can significantly affect accuracy, underscoring models' dependence on contextual cues. This study investigates how context manipulation influences both model accuracy and feature attribution, providing insights into the reliance of object recognition models on contextual information as understood through the lens of feature attribution methods. We employ a range of feature attribution techniques to decipher the reliance of deep neural networks on context in object recognition tasks. Using the ImageNet-9 and our curated ImageNet-CS datasets, we conduct experiments to evaluate the impact of contextual variations, analyzed through feature attribution methods. Our findings reveal several key insights: (a) Correctly classified images predominantly emphasize object volume attribution over context volume attribution. (b) The dependence on context remains relatively stable across different context modifications, irrespective of classification accuracy. (c) Context change exerts a more pronounced effect on model performance than Context perturbations. (d) Surprisingly, context attribution in `no-information' scenarios is non-trivial. Our research moves beyond traditional methods by assessing the implications of broad-level modifications on object recognition, either in the object or its context.