Professional Documents
Culture Documents
1 Introduction
Robot navigation is the task where an autonomous robot moves safely from
one location to another. This involves three primary questions [Leonard and
Durrant-Whyte, 1991]:
• Global positioning - The robot is initially lost. i.e. can deal with
multiple ideas (hypotheses) about its location.
1
• kidnapped robot problem - The robot is suddenly it is ’kidnapped’ to a
new location. Here the robot needs to determine that it has been kid-
napped and, furthermore, must work out its new location. The global
positioning problem is a special case of the kidnapped robot problem
where the robot is told it has been kidnapped [Fox et al., 1999].
3 Localisation Algorithm
The Monte Carlo Localistion is implemented via the approximation of bayes
rule [Couch, 2001]:
4. Resample new particles from the existing particle set according to the
weights. Replace the particle set with this re sampled set. Go to 1.
2
3.2 Perceptual Model
The perceptual model weights a given pose via the use of a sensor, which is
this projeect was a N ring sonar sensor.
For this project sonar sensor weight for is determined by:
X
N
wi = 1 − (mean − xi )2 /N/ρ (2)
k=0
3.3 Resampling
This project used sample/importance resampling (SIR) to resample the set of
particles. The threshold used to test the weight for each sample was chosen to
be 1/particle count. Furthermove as an extension on the project, the number
of particles will also be updated depending on the following condition:
particle_count = PARTICLE_COUNT_MAX:
InitaliseUniformPose(particle_count)
IncreaseParticleCount(&particle_count);
DecreaseParticleCount(&particle_count);
return particle_count;
}
3
References
[Couch, 2001] L. Couch. Digital and Analog Communication Systems. Pren-
tice Hall, 2001.
[Fox et al., 1999] D. Fox, W. Burgard, and S. Thrun. Markov localization for
mobile robots in dynamic environments. Journal of Artificial Intelligence
Research, 11, 1999.
[Leonard and Durrant-Whyte, 1991] J. Leonard and H. Durrant-Whyte.
Mobile robot localization by tracking geometric beacons. IEEE Trans-
actions on Robotics and Automation, 1991.
[Thrun et al., 2000] S. Thrun, D. Fox, W. Burgard, and F. Dellaert. Robust
monte carlo localization for mobile robots. Artificial Intelligence, 128(1-
2):99–141, 2000.