This study introduces an active inspection system. It dynamically selects camera viewpoints to reveal defects, using real-time texture inconsistency mapping with reflectivity compensation for accurate assessment on glossy or metallic surfaces.

Harshith Rohan Iyyer, Phoebe Lynn Swanson, Zhang Rui Xiang, Kenzo Masaki Fujimura, Beatrice Elise Dawson

This paper presents a visual grasping system. Depth and semantic segmentation guide affordance-based grasp point prediction for mobile robots operating in cluttered or unfamiliar indoor environments, improving reliability in autonomous manipulation tasks.

Rudra Ajeet Bharadwaj, Lydia Jane Fletcher, Sun Wei Long, Takahiro Naoki Yamashita, Clarissa June Middleton

This paper proposes a forecasting framework for pedestrian trajectories. Using temporal graph neural networks with occlusion-aware embedding, it predicts motion flow accurately in crowded urban intersections to assist autonomous vehicle navigation.

Samar Krish Dev, Angela Brooke Harmon, Liu Zhao Xing, Riku Hideki Nakamoto, Georgia Mae Delaney

This study introduces a driver behavior recognition system. By analyzing pose dynamics and gaze direction, it detects lane change intentions in real time, improving proactive decision support for advanced driving assistance systems.

Aryan Neel Desai, Melissa Skye Thornton, Zhao Hui Ming, Nobuaki Koji Yamazaki, Eleanor Kate Stafford

This paper introduces an unsupervised learning approach to discover novel objects. By clustering visual concepts through interactive episodes, robots autonomously expand their perception capabilities for unknown entities in exploration and manipulation tasks.

Rahil Nishant Malhotra, Charlotte Hazel Donovan, Wu Liang Zhi, Junya Takeshi Kobayashi, Amelia Faith Burton

This research presents a scene graph system for robotic manipulation. It models object relationships through context-aware graph construction, enabling robots to reason about object roles and spatial dynamics in cluttered or semantically rich indoor environments.

Aarav Ishaan Naidu, Evangeline Brooke Pierce, Huang Wen Zao, Ryota Shinji Nakamura, Tabitha Claire Daniels

This study introduces a transformer-based architecture for understanding crowd behaviors. It models spatial and temporal dependencies, identifying anomalous activities in real time to support autonomous monitoring in smart city surveillance deployments.

Rudraksh Deep Saini, Georgia Lynn Walters, Zhang Bo Rui, Kohei Satoshi Fujikawa, Imogen Felicity Doyle