This paper proposes a correlation-guided method for accurate scene flow estimation in dynamic environments. The model refines 3D motion fields in real-time, helping autonomous systems anticipate movement and plan trajectories with improved responsiveness.

Chen Hao Sheng, Liu Wen Jian, Xu Tian Hui, Zhang Ming Bo, Gao Rui Cheng

This research introduces a domain adaptation pipeline for all-weather perception. It employs GAN-based image translation and contrastive loss to generalize visual models across fog, rain, and night conditions without labeled target data.

Jessica Monroe, Eric Patterson, Natalie Rivers, Gregory Armstrong, Victoria Hughes

This paper introduces a cross-view matching framework for aligning aerial and ground perspectives in autonomous navigation. The system enhances geolocation accuracy using deep visual embedding and spatial alignment techniques across heterogeneous vision sources.

Chen Wen Rui, Liu Tian Hao, Xu Zhi Lin, Zhang Rui Cheng, Gao Bo Liang

This study presents a multi-agent tracking model that integrates temporal memory with spatial refinement. It maintains object continuity and accuracy in high-speed autonomous systems such as drones, delivery robots, and urban mobile platforms.

Emma Donovan, Lucas Barrett, Zoe McAllister, Noah Clark, Eleanor Jameson

This research introduces a visual SLAM system enhanced with attention-based modules. It improves pose estimation and map consistency in low-feature areas, supporting indoor navigation for autonomous cleaning, inspection, and warehouse robotics.

Chen Hao Jie, Zhang Fang Wei, Liu Rong Jian, Xu Lin Tao, Gao Xian Rui

This paper proposes a robust lane detection model that aggregates temporal and multi-scale features. It excels under occlusion and degraded visibility, enhancing navigation safety in autonomous road systems.

Charlotte Hayes, William Dorsey, Amanda Riley, Benjamin Griffin, Natalie Freeman

This work introduces a domain adaptation strategy for autonomous driving. It adapts models trained in one season to operate effectively in others using image translation and feature consistency learning, eliminating retraining needs.

Chen Yu Zhen, Liu Hao Sheng, Xu Min Qiang, Zhang Bo Tian, Gao Liang Rui