Page - 198 - in Proceedings - OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“
Image of the Page - 198 -
Text of the Page - 198 -
Toavoidunintendedmovementsof thekingpin joints, theGazebosreal timeupdatehas tobe2000Hz
and themaximum step size0.0005s.
The following three improvements would increase the accuracy of the simulation. Firstly, detailed
measurements should be taken to replace the wheels approximated friction parameter. Secondly,
damping should be added to the vehicle. Finally, the front wheels should be powered and equipped
withadifferential.
7. Results
Two tests are carried out to quantify the accuracy of the real and the simulated robots motion. For
thefirst test, a semicirclewith themaximumsteeringangleandavelocityof0.1m/swasdriven. The
low speed used during the test allows for errors stemming from wheel slipping and centrifugal force
to be ignored. The motion model represents the motion commands in this test, so it can be used as a
reference. The radius of the semicircle driven by the real car is 5cmbigger than the reference. This
is caused by the unsteady steering of the RC-race car. The simulated car drives a trajectory differing
fromacircle. During thewhole test, thepositionsof the realand thesimulatedcararecoveredby the
covarianceellipse. InFigure5a the test results are shown.
For the second test, a straight linewas drivenwith avelocity of0.1m/s, basedon themotion models
response. The real car stops4.5cmbefore the reference, because of inaccuracies in the measurement
of the wheel size. The simulated car stops 1.8cmbehind the reference. The reason for this deviation
is that unlike Gazebo, the motion model does not regard the kinetic energy of the vehicle. Again, the
covarianceellipsecovers thepositionof the realand thesimulatedvehicle. The test resultsareshown
inFigure5b.
To increase the accuracy of the motion model, two improvements can be made. Firstly, the number
of updates can be increased to downsize the time steps. Secondly, the kinetic energy of the vehicle
shouldbeconsideredby thevelocitymotion model.
MotionModel
RealCar
SimulatesCar
Covariance
(a) MotionModel
RealCar
SimulatesCar
Covariance
(b)
Figure5: Comparison of trajectory and visualisation of the (a) rotational and (b) the straight behavior
of thecovarianceellipse.
198
Proceedings
OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“
- Title
- Proceedings
- Subtitle
- OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“
- Authors
- Peter M. Roth
- Kurt Niel
- Publisher
- Verlag der Technischen Universität Graz
- Location
- Wels
- Date
- 2017
- Language
- English
- License
- CC BY 4.0
- ISBN
- 978-3-85125-527-0
- Size
- 21.0 x 29.7 cm
- Pages
- 248
- Keywords
- Tagungsband
- Categories
- International
- Tagungsbände