UWE Bristol Engineering showcase 2015
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Ji Qiu<br />
BEng (Hons) Electronic <strong>Engineering</strong><br />
Project Supervisor<br />
Professor Quan Min Zhu<br />
Application of Hidden Markov Model to Locate Soccer Robots<br />
Introduction<br />
The Hidden Markov Model is a statistical model with unobserved and independent states. This project aims to advance the applicable algorithms by Hidden<br />
Markov Model as a basis for predicting the probabilities in location of soccer robot’s trajectories. Moreover, the algorithm adopts Hidden Markov Model as a<br />
means of estimating and predicting position of soccer robot with given map and sensory data. Finally, computational experiment approach was used to<br />
demonstrate the analytical results with MATLAB simulations.<br />
Rolling-Back Algorithm<br />
As Hidden Markov Model shown in Figure 1, the dynamic process for the<br />
probability in total states can be described as below:<br />
<br />
pp 1 tt + ∆tt = ww 1,1 pp 1 tt + ww 2,1 pp 2 tt + ⋯ + ww nn,1 pp nn tt<br />
pp 2 tt + ∆tt = ww 1,2 pp 1 tt + ww 2,2 pp 2 tt + ⋯ + ww nn,2 pp nn tt<br />
…<br />
pp nn tt + ∆tt = ww 1,nn pp 1 tt + ww 2,nn pp 2 tt + ⋯ + ww nn,nn pp nn tt<br />
Where tt (0, 1, 2 … ) is the discrete time index. When tt = 0, pp 1 0 = 1,<br />
pp ii 0 = 0(ii = 2,3, … , nn)<br />
Application in Predicting<br />
Suppose that the original situation is like figure right hand side. Assume that<br />
up move is the direction to goal. That is, the connect probability from<br />
beginning state to up state is the largest one. Generally, left and right<br />
directions are the same value and down direction has the least probability.<br />
Blue and red rectangles mean different teams when the circle one represents<br />
the soccer robot which is analysed carrying the soccer ball. As other robots<br />
are obstacles, blue and red rectangles are unwalkable.<br />
Computational Application<br />
This simulation generally contains three<br />
parts as follows:<br />
• Field specification<br />
• Determine transition probability matrix<br />
• Position prediction<br />
This progress can be organised as an<br />
excellent self-adaptation control system.<br />
The rolling-back algorithm defaults that the<br />
Markov Model discussed here satisfies<br />
homogeneity. But in actual practical<br />
application, the stochastic sequence may<br />
not satisfy the homogeneity. Under this<br />
situation, perhaps overlying the predicting<br />
probabilities of different orders should be<br />
considered together, because these<br />
probabilities may donate to the prediction.<br />
Furthermore, the influences of these<br />
probabilities of different orders on the<br />
prediction can be different in fact. In order<br />
to improve the system, this kind of<br />
considerations should be taken into<br />
account. That is to say, these predicted<br />
results of different orders should be<br />
superimposed with different weighting<br />
factors.<br />
Assume that the probability of initial state of<br />
the system can be represented by a vector PP ss .<br />
After ∆tt = 1 times, the relationship between<br />
PP ss and the expected probability vector<br />
PP ss (tt + 1) = (aa 1 , aa 2 , aa 3 … aa nn ) could be<br />
described as the equation (Rolling-Back<br />
Algorithm)below:<br />
PP tttttttttttttttttttt ∈ PP ss (tt) PP ss tt + 1 = PP ss tt WW mmmmmm<br />
Where WW is a matrix.<br />
• Step 1: Calculate the probability by Hidden Markov Model<br />
Algorithm.<br />
• Step 2: Find the best trajectory by compare these<br />
probabilities above.<br />
• Step 3: The robot could make a decision for next step.<br />
Hence, the available state should be detected again at the<br />
same time and calculate the initial probabilities originally<br />
from present state to next step in expected state.<br />
Figure 3: Example simulation<br />
Figure 1: Markov Chain<br />
Figure 2: Simple example<br />
Project summary<br />
This project adopts a Hidden Markov Model as a basis<br />
for predicting the probabilities in location of soccer<br />
robot’s trajectories, and uses computational<br />
experiment approach to demonstrate the analytical<br />
results with MATLAB simulations.<br />
Project Objectives<br />
• Take critical survey on the up-to-date status of the<br />
relevant research and applications.<br />
• Tailor the theory of Markov chains into a<br />
condensed pack for the problems.<br />
• Select parameter estimation algorithms.<br />
• With MATLAB, design program to implement the<br />
numerical algorithms with initial demonstration<br />
of the computational effectiveness and accuracy.<br />
• Apply the Markov model and its parameter<br />
estimation algorithm to soccer robot’s localisation<br />
for further bench test of the theoretical results.<br />
• Design user friendly manual to run the<br />
demonstration programs.<br />
Project Conclusion<br />
This project has formulated a research problem<br />
(location of soccer robots), proposed a solution<br />
procedure (based on Hidden Markov Model), and<br />
finally used computational experiment approach to<br />
demonstrate the analytical results with MATLAB<br />
simulations. With the research understanding,<br />
Hidden Markov Model has been a very useful<br />
framework to accommodate predicting problems<br />
encountered in many stochastic systems, which its<br />
applications in soccer robot location could promote<br />
many other interesting research issues as well as<br />
increasing the robot intelligence. Once again this is<br />
the beginning stage presented in the paper and<br />
comprehensive future studies have been planned<br />
accordingly.