Out of sequence measurement in multi sensor fusion

Abstract

Autonomous vehicles perceive the surrounding environment by analyzing and correlating the data received from multiple sensors. The detection ability and accuracy are constrained by environmental conditions. These constraints can be overcome by fusing data from multiple sensors. But in multi-sensor fusion, Out of Sequence Measurement (OOSM) problem is increasing. In such a system, information is received at differed time at the fusion center due to different processing times of the sensors, cycle time, and communication delay which causes the later measurement from the higher rate sensor to overtake the newer measurement from the lower rate sensor. Out of sequence measurement problem can be avoided by introducing a delay in a higher rate sensor. A fixed lag algorithm is used which helps to synchronize the data at the fusion center by introducing a delay. Another approach for solving out of sequence is by predicting the data from the lower rate sensor when data from the higher rate sensor is available is done using regression. The algorithms are implemented on real-time data and the results of the fixed lag smoothing had delay hence the second approach is adaptable which will have no delay infusion.