Signal Processing

1
Signal Processing

• Jeremy Wyatt
• Intelligent Robotics
• School of Computer Science

2
Sonar

• Sound wave transmitter/receiver
• Sound wave bounces back off objects
• Counter measures time of flight
• Calculates the distance assuming speed of sound
  is 335ms-1
• Errors due to specular reflection

3
Typical data

• We gathered 100 samples from a sonar in three
  positions relative to a wall
• You get a figure for time of flight (T) in units
  of 0.5 microseconds
• To convert to metres

4
(No Transcript)
5
(No Transcript)
6
Median and mean

• Order the values by size
• take the central value, here 1.425
• mean is 1.458
• median is more robust to outliers

1.41, 1.38, 1.71, 1.45, 1.42, 1.39, 1.43, 1.47
1.38, 1.39, 1.41, 1.42, 1.43, 1.45, 1.47, 1.71
7
Mapping

• We took 25 sonar readings each at 100 equally
  spaced angles between 0 and 360 degrees

8
Plot range versus angle
Range in metres Angle in degrees
9
Convert to Cartesian frame
x
y
r
q
10
(x,y)
r

• q

r
11
Finding Walls

• Assume walls are straight lines
• Each point is on many possible lines
• Each line can be described by an equation

12
(No Transcript)
13
Hough Transform

• Create f and w for all possible lines
• Create an array A indexed by f and w
• for each point (x,y)
• for each angle f
• w xcos(f) ysin(f)
• Af,w Af,w1
• end
• end
• where A gt Threshold return a line

14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
Summary

• Use median
• Hough finds lines
• But because of width of sonar beam gives bad
  results
• May be better with Infra Red
• Smooth in Hough Space
• Use property of sonar

19
(No Transcript)