Temporal Logic Planning

a) Bob's garden. A bee robot is tasked with pollinating the flowers. The environment is uncertain due to the presence of birds, varying weather conditions, and the dynamics of the robot. b) Bob has a preference for how the robot should achieve the task of pollination. The relation defining Bob's preference is a partial order rather than a total order. While planning with preferences forming a total order relation has been well-studied, the question arises: how do we rank plans when dealing with user preferences that form partial order relations? And consequently, how should the planning be conducted?

Description

Preferences play an important role in guiding decision-making in situations where a single goal can be achieved in multiple ways or a set of goals are conflicting, meaning that not all of them can be attained simultaneously. In these situations, the robot can ask the user to provide their preference on how to achieve the goal or their preference over multiple goals. However, this might not always be possible, especially in stochastic and dynamic environments where the human operator is unavailable, or the robot needs to make decisions and action plans in a timely manner. This project aims to provide algorithms and theories for planning and decision-making in stochastic and dynamic environments.

References

Hazhar Rahmani, Abhishek N Kulkarni, Jie Fu. Probabilistic Planning with Partially Ordered Preferences over Temporal Goals. In Proc. IEEE International Conference on Robotics and Automation, 2023.
Lening Li, Hazhar Rahmani , Jie Fu. Probabilistic Planning with Prioritized Preferences over Temporal Logic Objectives. In Proc. International Joint Conference on Artificial Intelligence, 2023.
Hazhar Rahmani, Abhishek N Kulkarni, Jie Fu. Preference-Based Planning in Stochastic Environments: From Partially-Ordered Temporal Goals to Most Preferred Policies. Preprint, 2024.

Hazhar's home page