矩阵式函数迭代惯导解算的FPGA实现

The implementation of FPGA from Matrix-Form Functional Iteration for Inertial Navigation Computation

惯性导航解算的目的是处理陀螺仪和加速度计的输出信号,以获取运动物体的姿态、速度和位置信息。函数迭代惯导解算方法(iNavFIter)将运动不可交换性误差降低到了计算机截断误差水平,但计算负担是经典双子样算法的10倍左右。为了提高计算效率,提出了iNavFIter-M算法,将函数迭代惯导解算重新编排为矩阵向量形式,并在硬件FPGA上实现了iNavFIter-M算法的IP核。数值结果表明:所设计的iNavFIter-M算法IP核实现了1000Hz采样率下的高精度实时计算。

The purpose of inertial navigation solution is to process the output signals of gyroscope and accelerometer to obtain the attitude,velocity and position information of moving objects.The functional iteration algorithm(iNavFIter)reduces the motion noncommutativity errors to the level of computer truncation error,but the computational burden is about ten times more than that of the classical two-sample algorithm.In order to improve the computational efficiency,the iNavFIter-M algorithm proposed recently rearranges the functional iterative inertial navigation solution into a matrix-vector form.The IP core of iNavFIter-M algorithm is implemented on the hardware FPGA.The numerical results show that the IP core of iNavFIter-M algorithm designed realizes high-precision real-time computation at a sampling rate of 1000Hz.