Full text
| |
| Author(s): |
Luis Gustavo Rocha Vianna
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Instituto de Matemática e Estatística (IME/SBI) |
| Defense date: | 2015-05-18 |
| Advisor: | Leliane Nunes de Barros |
| Abstract | |
This work is a study on the planning problem in artificial intelligence, specifically probabilis- tic planning in continuous spaces. The efficient solution of planning problems is a major goal in artificial intelligence and can be applied extensively in autonomous agents. In many applications, the modelled problem contains continuous resources, so that an optimal planner must reason over continuos quantities to obtain appropriate actions. A recent and exact solution is Symbolic Dyna- mic Programming, which extends discrete probabilistic planning solutions to continuous problems by using a symbolic representation of state variables. This solution is interesting because it can find optimal solutions, however it is limited in efficiency because it relies on standard dynamic pro- gramming and doesn\2019t use initial state information or heuristic search. On this work, I will extend Symbolic Dynamic Programming to use more efficient dynamic programming approaches, based on recent solutions for discrete probabilistic planning. A novel planner using symbolic representation and heuristic search is proposed and compared to previous works on relevant continuos scenarios. (AU) | |
| FAPESP's process: | 11/16962-0 - Real Time Dynamic Programming and Monte-Carlo Simulation for Probabilistic Planning |
| Grantee: | Luis Gustavo Rocha Vianna |
| Support Opportunities: | Scholarships in Brazil - Master |