基于联邦滤波的智能车辆多传感器组合导航的研究

Research on Multi-Sensor Integrated Navigation Technique Based on Federated Filter for Intelligent Vehicle

为满足复杂环境下自主驾驶智能车辆的高可靠性导航要求,对智能车辆的多传感器组合导航进行了研究。提出了一种基于自适应联邦卡尔曼滤波的智能车辆SINS/CP-DGPS/双向光电测速仪组合导航方法,根据联邦滤波的分散滤波结构,分别建立了各滤波器的模型,进行了仿真试验验证。结果表明,该组合导航系统能为智能车辆提供丰富的导航信息,具有100Hz的高频输出、厘米级的导航精度和较强的容错能力,便于实现对发生故障的传感器的隔离。当GPS较长时间中断时,组合系统通过SINS/光电测速仪所构成的局部滤波器的辅助仍能为智能车辆提供可靠的导航数据。

To satisfy the high--reliable navigation demand in the complex environment for the autonomous intelligent vehicle, a new multi--sensor integrated navigation technique was studied. An integrated SINS/CP--DGPS/speedometer navigation method based on the self--adaptive federated Kalman filter was proposed. According to the separate configuration of federated filter, several filter models of this navigation method were then set up. Finally, simulation experiments were made. Simulation results show that the proposed navigation method can provide enough navigation information at 100Hz with cm--level navigation accuracy and good fault--tolerant ability, which is very convenient to separate the fault sensor from the integrated system. Even when GPS is continuously interrupted for a period, the intelligent vehicle can still get reliable navigation information by the aid of the local filter formed by SINS/Speedometer.