Jane Jean Kiam (UniBw M)
With digitization, more information is made available, even in real time, to improve decision making. However, the abundance of information complicates the decision-making process too. In this talk, I will share several real-world applications that we are working on, in which continuous automated decision-making support is essential. Typically, applications involving in-cockpit assistance, as well as operations of multiple unmanned vehicles (including manned-unmanned teaming) for disaster response, require real-time decision-making support, and are the two focuses of this talk. Subsequently, I will discuss some of the “polytheist” solution frameworks we explore, each made up of various AI-methods, ranging from (multi-agent) motion/path planning, task and motion planning, human-aware hierarchical task network planning, probabilistic sequential decision making for reactive planning, to planning and acting. Some typical challenges faced with regards to the implementation and the transferability will also be outlined.