ESKF situation #51
Description
Excuse me, how do we calculate the input features in ESKF (Error State Kalman Filter)? Generally, the error state is reset after measurement update. The next time fw_evol_diff and fw_update_diff are input to the neural network, they will be 0, missing information. How can we do?
怎么选取eskf框架下的特征?fw_evol和fw_update都选名义状态的可以吗?