THIS GRANT IS A CONTINUATION OF N00014-13-1-0778 Perception-Based, Reactive, Temporal-Logic Planning for Autonomous Deck Operations
Abstract
Year 1 Perception work in year 1 will focus on establishing the foundation for imperfect information about static objects. The reactive synthesis work will focus on Thrust III, i.e., creating a specification language, problem statement, and synthesis algorithms that account for the dual objectives of information gathering and mission progress. We will create the initial version of the case study with the typical constraints and specifications but possibly with complete sensing. Year 2 Perception work in year 2 will focus on the spatial and temporal consistency of semantics extracted by perception as well as on the semantics of dynamic objects. The synthesis effort will move to Thrust IV to incorporate probabilistic models of perception uncertainty as well as to the investigation of the causes for potential high complexity along with identification of possible system-theoretic and/or problem-specific structures that may help reduce this complexity (i.e., Thrust V). The case study will be modified to capture the uncertainty models that have been developed and the features of synthesis that we have been investigated so far. Year 3 focuses on the integration of imperfect perception and on an extensive evaluation with a variety of vehicle models. Beyond usual recognition accuracy rates, a metric for evaluation will be the comparison in planning performance between a deterministic scenario where perception gives certain but sometimes erroneous semantics vs a probabilistic scenario with all information from perception being imperfect. The focus in synthesis work will move to coping with the complexity."
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 10, 2016
- Source ID
- N000141612051
Entities
People
- Ufuk Topcu
Organizations
- Office of Naval Research
- United States Navy
- University of Pennsylvania