Aiming at problem that navigation errors of traditional vehicular dead reckoning（ DR） are accumulated rapidly with time

a low-cost vehicular DR system based on multi-sensor fusion is proposed. Use unscented Kalman filter（ UKF） as data fusion algorithm

information from multiple sensors

including odometer

gyroscope

accelerometer

magnetometer and barometer

are fused to restrain accumulated errors of DR system

realize highprecise navigation using low cost

low precision sensor. Through real road navigation experiment for 12 min

result shows that navigation error is less than 1 % of the total distance

the precision is hight improved compared with traditional DR system.