FLiCR: A Fast and Lightweight LiDAR Point Cloud Compression Based on Lossy RI

Abstract

Light detection and ranging (LiDAR) sensors are becoming available on modern mobile devices and provide a 3D sensing capability. This new capability is beneficial for perceptions in various use cases, but it is challenging for resourceconstrained mobile devices to use the perceptions in real-time because of their high computational complexity. In this context, edge computing can be used to enable LiDAR online perceptions, but offloading the perceptions on the edge server requires a lowlatency, lightweight, and efficient compression due to the large volume of LiDAR point clouds data.This paper presents FLiCR, a fast and lightweight LiDAR point cloud compression method for enabling edge-assisted online perceptions. FLiCR is based on range images (RI) as an intermediate representation (IR), and dictionary coding for compressing RIs. FLiCR achieves its benefits by leveraging lossy RIs, and we show the efficiency of bytestream compression is largely improved with quantization and subsampling. In addition, we identify the limitation of current quality metrics for presenting the entropy of a point cloud, and introduce a new metric that reflects both point-wise and entropy-wise qualities for lossy IRs. The evaluation results show FLiCR is more suitable for edge-assisted real-time perceptions than the existing LiDAR compressions, and we demonstrate the effectiveness of our compression and metric with the evaluations on 3D object detection and LiDAR SLAM.

Poster: Making Edge-assisted LiDAR Perceptions Robust to Lossy Point Cloud Compression

“…Lossy compression methods achieve a higher compression ratio than lossless compression by reducing the level of detail (LoD) at the IR level and losing points in the point cloud. Among the IRs, the range image (RI) has a low-latency benefit with the simplicity of its conversion process [19] as it is generated by converting the points in the 3D Cartesian coordinates to the spherical coordinates and mapping the converted points into a 2D image. Each pixel of a RI has a depth value.…”

“…Additionally, the PCC method should be lightweight to operate on mobile devices and low latency for the latency-performance tradeoff of real-time perceptions; the higher end-to-end latency causes larger discrepancies between the perception result and the real-world environment [14]. In our recent work, we showed that existing PCC methods [15]- [18] are hardly suitable for remote real-time perceptions and presented a fast and lightweight PCC method, FLiCR [19].…”

A Survey on Data Compression Techniques for Automotive LiDAR Point Clouds

“…The method uses the H.264 video codec, a video standard that uses a motion-compensated inter-frame prediction, intra-frame prediction, transformation and quantization, and entropy coding. Similarly, Heo et al [79] further explore this concept by applying the H.264 video codec to lossy RIs with the primary goal of achieving low latency results. Using a different approach, Nardo et al [39] applied the LZW [80] algorithm and MJ2 [60] to a stream of RIs.…”