关键词: 关键词: 室内地图 地图更新方法 分层式更新 卷积神经网络

Abstract: Abstract: The existing map updating method is not ideal under the indoor map environment. In view of this situation, this paper proposes a layered indoor map updating method. It uses the activity of indoor objects as a parameter to divide the levels to reduce the amount of updated data, and uses Convolution to determine the network's level of ownership of the indoor data. The experimental results show that compared with the incremental update method and the version update method without layering, the method proposed in this paper has greatly improved both in the update time and the update data amount.

Key words: Keywords: indoor map map update method tiered update convolutional neural network