多目标多传感器数据融合轨迹追踪的建模与仿真

Multi-target multi-sensor data fusion's modelling and simulation for trajectory tracking

利用卡尔曼滤波(KF)和交互多模型(IMM)滤波进行传感器多目标测量的建模,在多目标轨迹跟踪中利用JPDAF和神经网络融合算法。轨迹联合和数据融合用于融合轨迹数据时假设,2只传感器追踪单目标到3只传感器追踪3个目标,在此基础上评估了多个散布式传感器对于单目标、双目标和多目标的测量效能。对于不同滤波器的性能进行了比较,并得到轨迹融合可以很好地逼近真实轨线,优于其他任何传感器目标测量方法。

An executing of track fusion using various algorithms has been demonstrated. The sensor measurements of these targets are modeled using Kalman filter （KF） and interacting multiple models （ IMM ） filter. The joint probabilistic data association filter （JPDAF） and neural network fusion （NNF） algorithms were used for targets tracking. Track association and fusion algorithm are executed to get the fused track data for various scenarios ,from two sensors tracking a single target to three sensors tracking three targets, to evaluate the effects of multiple and dispersed sensors for single target, two targets, and multiple targets. The targets chosen were distantly spaced, closely spaced and crossing. Performance of different filters was compared and fused trajectory is found to be closer to the true target trajectory as compared to that for any of the sensor measurements of that target.