Aykut Erdem
Papers
4
Total Citations
33
H-Index
3
About
Aykut Erdem is a researcher whose work spans computer vision, machine learning, and robotics, with a particular focus on developing intelligent systems capable of understanding and interacting with complex visual environments. His research demonstrates a consistent interest in bridging perception and physical reasoning, tackling challenges that arise at the intersection of these domains. One of Erdem's notable contributions involves advancing person tracking under difficult real-world conditions. His 2021 work on synthetic data generation for tracking under adverse weather conditions — garnering 17 citations — addresses a critical bottleneck in training robust visual systems: the scarcity of labeled real-world data in challenging environments. By leveraging synthetic data, his approach offers a scalable pathway to more resilient tracking systems. Erdem has also made meaningful strides in robotic manipulation, particularly in push effect prediction. His recurring investigations into object- and relation-centric representations for predicting the outcomes of push actions reflect a deep engagement with non-prehensile manipulation — a foundational skill in robotics spanning pre-grasp planning to scene rearrangement. This body of work, accumulating citations across multiple iterations from 2021 to 2024, underscores both its growing relevance and Erdem's commitment to refining physically grounded, relational reasoning frameworks for autonomous agents.
Research Focus
Key Achievements
Top Papers
- 1Using synthetic data for person tracking under adverse weather conditions17 citations · 2021
- 2Object and relation centric representations for push effect prediction9 citations · 2024
- 3Object and Relation Centric Representations for Push Effect Prediction4 citations · 2023
- 4Object and Relation Centric Representations for Push Effect Prediction3 citations · 2021