LiDAR-based localization using universal encoding and memory-aware regression

Highlights：

• A novel LiDAR localization framework that performs absolute pose regression (APR) with novel universal encoding to avoid redundant retraining of the whole network from scratch and preserve the privacy of training data.

• A memory regressor for memory-aware regression where the hidden unit numbers in the regressor determine the memorization capacity. It can be used to derive and improve the upper bound of the capacity, adapting different memorization capacity requirements for different scene sizes.

• Experiments on both outdoor and indoor datasets demonstrate the effectiveness of the proposed framework, which outperforms state-of-the-art APR methods.