基于非线性优化双目 VIO 的在线时间偏差标定实现方法

The error accumulation problem in srereo visual-inertial odometry (VIO) systems based on nonlinear optimization is serious when operating for extended periods in low-texture environments. Therefore, we propose an online temporal calibration method for the stereo VIO system based on nonlinear optimization. This approach makes full use of the benefits of stereo cameras by constructing error factors using epipolar constraints, which enhances system robustness and state estimation accuracy while lessening the detrimental effect of feature point mismatches on time offset calibration. It is suitable for low-cost, self-assembled systems. Experiments on public datasets show that the proposed calibration method has higher accuracy and faster convergence speed than current advanced calibration methods, thereby improving the accuracy and robustness of system state estimation. Experiments in real-world scenarios also validate the effectiveness of the proposed method.