Mapping dynamic environments using Markov random field models
This paper focuses on dynamic environments for mobile robots and proposes a new mapping method combining hidden Markov models (HMMs) and Markov random fields (MRFs). Grid cells are used to represent the dynamic environment. The state change of every grid cell is modelled by an HMM with an unknown tr...
| Autor principal: | |
|---|---|
| Outros Autores: | , |
| Formato: | article |
| Idioma: | eng |
| Publicado em: |
2020
|
| Assuntos: | |
| Texto completo: | http://hdl.handle.net/10174/27573 |
| País: | Portugal |
| Oai: | oai:dspace.uevora.pt:10174/27573 |
| Resumo: | This paper focuses on dynamic environments for mobile robots and proposes a new mapping method combining hidden Markov models (HMMs) and Markov random fields (MRFs). Grid cells are used to represent the dynamic environment. The state change of every grid cell is modelled by an HMM with an unknown transition matrix. MRFs are applied to consider the dependence between different transition matrices. The unknown parameters are learnt from not only the corresponding observations but also its neighbours. Given the dependence, parameter maps are smooth. Expectation maximization (EM) is applied to obtain the best parameters from observations. Finally, a simulation is done to evaluate the proposed method. |
|---|