IMU的轨迹解算和航迹显示

最新推荐文章于 2024-09-14 18:04:44 发布

ego_vehicle

最新推荐文章于 2024-09-14 18:04:44 发布

阅读量4k

点赞数 2

文章标签： c++ 自动驾驶算法

本文链接：https://blog.csdn.net/qq_53576285/article/details/129647985

版权

基于ros操作系统，调用IMU数据包，利用数据解算小车运动的轨迹，并在rviz中实现轨迹的可视化。其中IMU四元数对于位移速度和加速度的转换。

轨迹解算和换机显示的代码：

//IMU航迹推算
#include <ros/ros.h>
#include <sensor_msgs/Imu.h>
#include <tf/transform_datatypes.h>
#include <tf/tf.h>//关于坐标系解算的一个消息库
#include "nav_msgs/Path.h"
#include "ctime"

using namespace std;

int imuReceivedNumber = 0;

//时间戳
double time_k, time_k_1, dt;
//重力加速度
const tf::Vector3 g(0,0,9.81);

tf::Vector3 acc_b_k, acc_b_k_1;
tf::Vector3 w_b_k, w_b_k_1;

//k+1时刻的线加速度,经过旋转及中值近似后
tf::Vector3 acc_k_1;
//k+1时刻的角速度,经过旋转及中值近似后
tf::Vector3 w_k_1;
//k及k+1时刻的速度
tf::Vector3 velo_k, velo_k_1;
//k及k+1时刻的位置
tf::Vector3 position_k, position_k_1;
//k及k+1时刻的姿态,用四元数来表示
tf::Quaternion q_k,q_k_1;

ros::Publisher pubdatadisplay;
nav_msgs::Path path;
//接受位置，实现可视化，并发布
void pubPathmsg(tf::Vector3 position_, tf::Quaternion quaternion_){
    geometry_msgs::PoseStamped display_imu;
    display_imu.header.stamp = ros::Time::now();
    display_imu.header.frame_id = "display";

    display_imu.pose.position.x = position_.getX();
    display_imu.pose.position.y = position_.getY();
    display_imu.pose.position.z = position_.getZ();

    geometry_msgs::Quaternion q_msgs;
    q_msgs.x = quaternion_.getX();
    q_msgs.y = quaternion_.getY();
    q_msgs.z = quaternion_.getZ();
    q_msgs.w = quaternion_.getW();
    display_imu.pose.orientation = q_msgs;

    path.poses.push_back(display_imu);
    path.header.stamp = ros::Time::now();
    path.header.frame_id = "display";
    pubdatadisplay.publish(path);
}

//接受imu消息，并对消息进行解算处理，得到推算航迹需要用到的信息
void imuHandler(const sensor_msgs::Imu::ConstPtr &imuIn){
    if(imuReceivedNumber == 0){
        acc_b_k[0] = imuIn->linear_acceleration.x;
        acc_b_k[1] = imuIn->linear_acceleration.y;
        acc_b_k[2] = imuIn->linear_acceleration.z;

        w_b_k[0] = imuIn->angular_velocity.x;
        w_b_k[1] = imuIn->angular_velocity.y;
        w_b_k[2] = imuIn->angular_velocity.z;

        time_k = imuIn->header.stamp.toSec();

        velo_k[0] = 0;
        velo_k[1] = 0;
        velo_k[2] = 0;

        position_k[0] = 0;
        position_k[1] = 0;
        position_k[2] = 0;

        q_k.setX(0);
        q_k.setY(0);
        q_k.setZ(0);
        q_k.setW(1);

        pubPathmsg(position_k, q_k);//接受k时刻IMU的消息
        imuReceivedNumber++;
    }else if(imuReceivedNumber == 1){
        acc_b_k_1[0] = imuIn->linear_acceleration.x;
        acc_b_k_1[1] = imuIn->linear_acceleration.y;
        acc_b_k_1[2] = imuIn->linear_acceleration.z;

        w_b_k_1[0] = imuIn->angular_velocity.x;
        w_b_k_1[1] = imuIn->angular_velocity.y;
        w_b_k_1[2] = imuIn->angular_velocity.z;

        time_k_1 = imuIn->header.stamp.toSec();

        dt = time_k_1 - time_k;

        w_k_1 = (w_b_k + w_b_k_1) / 2;
        tf::Quaternion dw;//姿态信息

        dw.setX(w_k_1[0] * dt);
        dw.setX(w_k_1[1] * dt);
        dw.setZ(w_k_1[2] * dt);
        dw.setW(1);//接收k+1时刻IMU的消息
//主要用了中值法来估计求得
        //求姿态
        q_k_1 = q_k * dw;
//求加速度时也需要知道姿态的信息
        //求加速度
        tf::Vector3 acc_b_k_r = acc_b_k.rotate(q_k.getAxis(), q_k.getAngle());
        tf::Vector3 acc_b_k_1_r = acc_b_k_1.rotate(q_k_1.getAxis(), q_k_1.getAngle());
        acc_k_1 = (acc_b_k_r - g + acc_b_k_1_r - g) / 2;

        //求位置，用的加速度平均值
    position_k_1 = position_k + velo_k * dt + (acc_k_1 * dt) * (acc_k_1 * dt) / 2;

    //求速度，同样用加速度平均值
    velo_k_1 = velo_k + acc_k_1 * dt;
        
//调用上面的可视化发布函数
    pubPathmsg(position_k_1, q_k_1);
//将k=1时刻的值变成下一个k时刻
    w_b_k_1 = w_b_k;
    acc_b_k = acc_b_k_1;
    time_k = time_k_1;
    q_k = q_k_1;
    position_k = position_k_1;
    velo_k = velo_k_1;
    cout << "dt: " << dt << " x: " << position_k_1[0] << " y: " << position_k_1[1] << " z: " << position_k_1[2] << endl;
}
}
//初始化节点并执行上面的回调函数
    int main(int argc, char **argv){
    ros::init(argc, argv, "scanRegistration");
    ros::NodeHandle nh;

    ros:: Subscriber subImu = nh.subscribe<sensor_msgs::Imu>("/imu/data_raw", 50, imuHandler);

    pubdatadisplay = nh.advertise<nav_msgs::Path>("/display", 100);

    ros::spin();

    return 0;
}

代码中有IMU的坐标变化，同时利用四元数求得加速度、速度和位移，其中平均加速度存在估计的成分。在每个时间步长循环中，体现了微积分估计的思想。关于四元数的理解和定义可以参考这一篇文章：

可视化理解四元数，愿你不再掉头发 - 知乎 (zhihu.com)