Research & Projects
Research & Projects
This research investigates grounded dialogue for human–robot interaction by integrating natural-language instructions with perceptual, spatial, temporal, and contextual information. Using the SCOUT dataset and the STAIT toolkit, the work focuses on linking dialogue with images, camera poses, annotated point clouds, and interaction history to improve the interpretation of ambiguous spatial references and complex instructions. The research also explores replay, visualization, and annotation-refinement capabilities to support multimodal data analysis and the development of structured inputs for future grounded dialogue and vision-language models.
The objective of this research is to improve human–robot communication by connecting natural-language instructions with perceptual, spatial, and contextual information. Using the SCOUT dataset and the STAIT toolkit, this work focuses on integrating dialogue with images, camera poses, annotated point clouds, and interaction history to reduce ambiguity, support annotation refinement, and enable more effective interpretation of complex human instructions by autonomous systems.
Exploring and surveying real-world scenarios using ground rover multi-robot systems (MRS) or multi-agent systems (MAS) necessitates the development and integration of strategies that enable autonomous performance, such as scalability and robustness. To support these tasks, it is crucial to incorporate parameters that address communication constraints, cybersecurity threats, and formation control.
The primary objectives in exploring and surveying real-world scenarios with ground rover multi-robot systems (MRS) or multi-agent systems (MAS) are centered on enabling effective and reliable autonomous performance. A key goal is achieving scalability, allowing the system to seamlessly expand and operate with larger numbers of robots without compromising efficiency. Equally important is ensuring robustness so that the system can function reliably in unpredictable and dynamic environments. To support these objectives, it is essential to address communication constraints, ensuring that coordination and data exchange remain efficient even under limited or unstable network conditions. Additionally, safeguarding against cybersecurity threats is critical to maintaining system integrity and preventing malicious disruptions. Finally, formation control plays a vital role in enabling structured and coordinated movement, which is necessary for efficient exploration and surveying tasks.