Ultrasonic localization using Monte Carlo estimator

Position of car-like mobile robot is estimated by Monte Carlo estimator (particle filter), using ultrasonic beacons and PRBS correlation. Playfiled is 3x2.1m (Eurobot'08), sampling 25Hz. Created by Eurobot team CTU Dragons at FEL ČVUT, http://dce.felk.cvut.cz/dragons/

There are 3 active beacons around the playfield, transmitting pseudorandom periodic binary sequence. Mixture of this is received by the robot and a delay is measured. From this measurement, the particle filter (Monte Carlo estimator) estimates likelihood of actual robot position. The probability density is displayed on this animation. 2D density in x, y is represented by a cloud inside, marginal densities are the yellow hills around. Beside of x and y, a time difference D is estimated (as the beacons and robot are not synchronized) and an oscillator drift D' is estimated as well.

For better quality, see animated GIF (7.5MB) http://duch.cz/mp/uzv/mcl-jizda1.gif

Category:

Science & Technology

Tags:

• particle
• filter
• Monte
• Carlo
• ultrasound
• localization
• navigation
• GPS
• estimator
• Kalman
• robot
• Eurobot
• FEL
• ČVUT