论文部分内容阅读
针对微型飞行器(Micro air vehicle,MAV)在室内飞行过程中无法获得GPS信号,而微型惯性单元(Inertial measurement unit,IMU)的陀螺仪和加速度计随机漂移误差较大,提出一种利用单目视觉估计微型飞行器位姿并构建室内环境的方法。在机载单目摄像机拍摄的序列图像中引入一种基于生物视觉的方法获得匹配特征点,并由五点算法获得帧间摄像机运动参数和特征点位置参数的初始解;利用平面关系将特征点的位置信息由三维降低到二维,给出一种局部优化方法求解摄像机运动参数和特征点位置参数的最大似然估计,提高位姿估计和环境构建的精度。最后通过扩展卡尔曼滤波方法融合IMU传感器和单目视觉测量信息解算出微型飞行器的位姿。实验结果表明,该方法能够实时可靠地估计微型飞行器的位置和姿态,构建的环境信息满足导航需求,适用于微型飞行器室内环境中的导航控制。
However, the gyroscope and accelerometer of the Inertial measurement unit (IMU) have a large random drift error. Therefore, a new method that uses monocular vision Estimating the pose of the micro-aircraft and building the indoor environment. A biometric-based method was introduced to sequence images captured by airborne monocular cameras to obtain matching feature points. An initial solution of motion parameters and position parameters of inter-camera was obtained by five-point algorithm. By using plane relationships, The position information of the camera is reduced from three dimensions to two dimensions. A local optimization method is proposed to solve the maximum likelihood estimation of the camera motion parameters and the location parameters of the feature points, which can improve the accuracy of pose estimation and environment construction. Finally, the pose of the MAV is calculated by extending the Kalman filtering method by integrating IMU sensors and monocular visual measurement information. The experimental results show that this method can estimate the position and attitude of the MAV in real time and reliably. The constructed environment information can meet the navigation requirements and is suitable for the navigation control in the MAV environment.