Iowa State Professors, Dr. Yan-Bin Jia, Dr. Bowen Weng, and Dr. Tichakorn Wongpiromsarn, will represent the Department of Computer Science at the International Conference on Robotics and Automation (ICRA) this May in Atlanta, Georgia. ICRA is the premier conference for robotics researchers.
Research Articles
Dr. Yan-Bin Jia, Professor, is first-author on the paper “Patch Tree: Exploiting the Gauss Map and Principal Component Analysis for Robotic Grasping” along with Iowa State Ph.D. graduate Yuechuan Xue (’22) and Iowa State Ph.D. student Ling Tang. The paper explores a method for grasp planning that uses differential geometry to handle the curved surfaces found in most everyday objects. The team proposes that by discretizing the surfaces into elementary patches based on Gaussian curvature, and then further refining them with principal component analysis, they can then create a path tree structure. This structure would help plan and optimize finger placements for grasping and tool manipulation.
Dr. Bowen Weng, Assistant Professor, is a contributor to “Enhancing Repeatability and Reliability of Accelerated Risk Assessment in Robot Testing,” which is accepted to ICRA 2025. The paper, written by Linda Capito and Guillermo Castillo, presents research which works to enhance existing accelerated testing frameworks through a new algorithm that provably integrated repeatability and reliability with the already established formality and efficiency seen in the existing framework.
Dr. Tichakorn (Nok) Wongpiromsarn, Assistant Professor, is first-author on the paper "Locally Homotopic Paths: Ensuring Consistent Paths in Hierarchical Path Planning.” Written alongside Marcelo Kallmann and Andreas Kolling, the joint work with Amazon Robotics discusses a system that helps a robot or vehicle follow a planned path while adjusting to changes in its surroundings. It introduces a hard constraint and a cost function to ensure that the adjustments remain consistent with the original path—such as avoiding obstacles on the same side—without unnecessarily requiring the adjusted path to stay close to the original. Experimental results show that this approach is better at handling errors regarding a robot or autonomous vehicle’s location compared to existing methods.
Additionally, Shashwata Mandal, a PhD student in Computer Science, is first author on the paper “Towards a Computationally-Efficient Framework for Multiagent Visual Surveillance” that has been accepted to ICRA. The paper was written alongside Dr. Sourabh Bhattacharya, a Courtesy Associate Professor in Computer Science and an Associate Professor, Mechanical Engineering at Iowa State University. In the paper, they propose a deployment problem for a team of mobile guards that try to visually track a team of mobile intruders. Such problems arise in security and surveillance problems. Through presenting several variations of proposed control laws that include capture and search, they demonstrate an improvement in the overall visual footprint of the team of mobile guards.
Workshop
Dr. Weng is also organizing a workshop at the conference, “Towards Reliable and Trustworthy Embodied AI in Everyday Scenarios.” The workshop is organized by Dr. Weng, Fan Shi (National University of Singapore), Siqi Zhou (Technical University of Munich), Mansur Maturidi Arief (Stanford University), Fadhil Ginting (Stanford University), Linda Capito (Transportation Research Center Inc.), and Shuo Feng (Tsinghua University).
The workshop will explore the reliability and trustworthiness of embodied AI systems in real-world deployments across a variety of everyday scenarios through bringing together a variety of tools and methodologies as well as highlighting current challenges, with the goal of advancing AI and robotic systems towards greater versatility and trustworthiness.
Note: a previous version of this article was published on February 21, 2025. The version published March 7, 2025 included the paragraph on Shashwata Mandal. No other information was changed.