SE(3)-Transformation on Neural Implicit Maps for Remapping Functions

Implicit representations are widely used for object reconstruction due to their efficiency and flexibility. In 2021, a novel structure named neural implicit map has been invented for incremental reconstruction. A neural implicit map alleviates the problem of inefficient memory cost of previous online 3D dense reconstruction while producing better quality.

However, the neural implicit map suffers the limitation that it does not support remapping as the frames of scans are encoded into a deep prior after generating the neural implicit map. This means, that neither this generation process is invertible, nor a deep prior is transformable. The non-remappable property makes it not possible to apply loop-closure techniques.

We present a neural implicit map based transformation algorithm to fill this gap. As our neural implicit map is transformable, our model supports remapping for this special map of latent features. With this function of transformation, our dense mapping model can well equiped into SLAM algorithm with BA and Loopclosure.

Our work is based on DI-Fusion's PLIVox. The transformation borrows SO3-equivariance encoding of local geometries.

We implement a remapping pipeline for 3D reconstruction. When pose-stream changes (because of BA or loop-closure), corresponding frames will be transformed and remapped.

With the poses stream from ORB-SLAM2, we generate this reconstruction of Replica datasets with our remapping function.

We additionally implement an attempt on KITTI-odometry 00 to verify the use of remapping. The quality of neural implicit mapping here still needs improvements.