Exploring AI and Autonomy for Unmanned Vehicles: Autonomous Gas Source Search and Localization Using Mobile Sensors in Uncertain Environments

Abstract

Unmanned vehicles are now pivotal in both military and civilian applications, significantly influencing various aspects of our daily lives. Among their many uses, the deployment of teams or swarms of unmanned vehicles is particularly noteworthy due to their adaptability, versatility, and collaborative potential in achieving shared objectives with inherent redundancy. The central research question driving my research is: how can multiple unmanned vehicles autonomously and efficiently accomplish assigned missions in uncertain and dynamic environments? Addressing this question requires a comprehensive approach that considers the dynamic constraints of unmanned vehicles, the integration of diverse sensors and information, and the complexities of operating in uncertain environments. Furthermore, the level of cooperation and decentralization among vehicles, particularly under limited computational and communication resources, needs to be carefully examined.

To achieve fully autonomous cooperative unmanned vehicle systems that require minimal human intervention while ensuring safety, timeliness, and efficiency, several key aspects of unmanned vehicles need to be addressed. These include: i) enhancing the level of autonomy through the development of high-level decision-making and planning algorithms under uncertain environments; ii) augmenting situational awareness with sensor/information fusion techniques, potentially integrating domain knowledge and learned models for safer operations and superior mission performance; and iii) seamlessly integrating decision-making, planning, and situational awareness to create a synergistic system for real world applications.

The autonomous cooperative unmanned systems will have broad applicability in defense, public, and industrial sectors, including search and rescue operations, environmental monitoring, agricultural robotics, infrastructure inspection, law enforcement, and logistics. This seminar will introduce the ongoing research efforts of the KAIST Field AI and Robotics (FAIR) on the above aspects while focusing on autonomous gas source search and localization using information-theoretic and leaning-based approaches for mobile sensors.

Speaker

Prof. Hyondong Oh is currently an Associate Professor at the Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea. He received his BSc and MSc degrees in Aerospace Engineering from KAIST in 2004 and 2010, respectively, and his PhD in Aerospace Engineering from Cranfield University, Cranfield, United Kingdom, in 2013. He was a Postdoctoral Researcher at the University of Surrey, Guildford, United Kingdom, from 2013 to 2014, a Lecturer at Loughborough University, Loughborough, United Kingdom, from 2014 to 2016, and an Associate Professor at the Ulsan National Institute of Science and Technology, Ulsan, Republic of Korea, from 2016 to 2025. His research interests include mission planning and task assignment, information-theoretic active perception, learning-based perception, planning and control, estimation and sensor/information fusion, and swarm and flocking control of multiple unmanned vehicles.