An Optimized FPGA-Based Real-Time NDT for 3D-LiDAR Localization in Smart Vehicles

Deng, Qiudong; Sun, Haofei; Chen, Fupeng; Shu, Yuhao; Wang, Hui; Ha, Yajun

doi:10.1109/tcsii.2021.3095764

Cited by 11 publications

(5 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Point cloud registration [14] is a commonly used relative localization method for mobile robots. The ICP (Iterative Closest Point) algorithm is the prevailing choice for point cloud registration, whose core idea is to minimize the Euclidean distance between corresponding points in the source and target point cloud maps through the least squares method [15].Another location algorithm based on point cloud registration is the NDT algorithm, whose core idea is to transform the point cloud map into a piecewise continuous probability density function in a two-dimensional space, then the current frame is rotated and translated to obtain the highest score for the data in the reference frame [16]. This transformed map is referred to as an NDT map, in which each map unit is modeled as a normal distribution, with its mean and covariance matrices representing the spatial geometric characteristics of the region.…”

Section: Metric Mapsmentioning

confidence: 99%

Map representations for mobile robot localization

Zhang,

2024

Fourth International Conference on Mechanical Engineering, Intelligent Manufacturing, and Automation Technology (MEMAT 2023)

View full text Add to dashboard Cite

Localization is a prerequisite for mobile robots to achieve path planning, autonomous navigation, and decision-making. In localization, maps provide mobile robots with the ability to perceive and model the environment, so the choice of an appropriate map representation has a significant impact on the localization effect of mobile robots. From the perspective of usage matching, this paper divides map representations for mobile robot localization into four categories based on different focuses of describing the environment, which are metric maps, feature maps, topological maps, and semantic maps. The principles, characteristics, and classic cases of these representations are also analyzed in this paper. Furthermore, the applicable working conditions of each type of map representation are summarized to provide effective guidance for mobile robot localization.

show abstract

Section: Metric Mapsmentioning

confidence: 99%

Map representations for mobile robot localization

Zhang,

2024

Fourth International Conference on Mechanical Engineering, Intelligent Manufacturing, and Automation Technology (MEMAT 2023)

View full text Add to dashboard Cite

show abstract

“…The authors of [48] proposed an implementation for semantic-assisted NDT-based scan registration on Xilinx ZCU102, and achieved a 35.85x speedup and a 14.3x better energy efficiency compared to ARM Cortex-A53. NDT optimizes a rigid transformation between LiDAR scan and map on a continuous search space in an iterative fashion, and requires a computation of probability density for all points in a scan.…”

Section: Related Workmentioning

confidence: 99%

A Universal LiDAR SLAM Accelerator System on Low-Cost FPGA

Sugiura

Matsutani

2022

IEEE Access

View full text Add to dashboard Cite

LiDAR (Light Detection and Ranging) SLAM (Simultaneous Localization and Mapping) serves as a basis for indoor cleaning, navigation, and many other useful applications in both industry and household. From a series of LiDAR scans, it constructs an accurate, globally consistent model of the environment and estimates a robot position inside it. SLAM is inherently computationally intensive; it is a challenging problem to realize a fast and reliable SLAM system on mobile robots with a limited processing capability. To overcome such hurdles, in this paper, we propose a universal, low-power, and resource-efficient accelerator design for 2D LiDAR SLAM targeting resource-limited FPGAs. As scan matching is at the heart of SLAM, the proposed accelerator consists of dedicated scan matching cores on the programmable logic part, and provides software interfaces to facilitate the use. Our accelerator can be integrated to various SLAM methods including the ROS (Robot Operating System)-based ones, and users can switch to a different method without modifying and re-synthesizing the logic part. We integrate the accelerator into three widelyused methods, i.e., scan matching, particle filter, and graph-based SLAM. We evaluate the design in terms of resource utilization, speed, and quality of output results using real-world datasets. Experiment results on a Pynq-Z2 board demonstrate that our design accelerates scan matching and loop-closure detection tasks by up to 14.84x and 18.92x, yielding 4.67x, 4.00x, and 4.06x overall performance improvement in the above methods, respectively. Our design enables the real-time performance while consuming only 2.4W and maintaining accuracy, which is comparable to the software counterparts and even the state-of-the-art methods.INDEX TERMS SLAM, scan matching, FPGA.

show abstract

“…The proposed accelerator combines a parallelized distance computation unit and a dedicated sorter unit to speed up the graph construction and NN search. Deng et al [20] present an FPGA-based accelerator for Normal Distributions Transform (NDT). NDT [21] models point clouds as a set of voxels, each of which represents the Gaussian distribution of points.…”

Section: B Fpga-based Acceleration For Point Cloud Registrationmentioning

confidence: 99%

An Efficient Accelerator for Deep Learning-based Point Cloud Registration on FPGAs

Sugiura¹,

Matsutani²

2022

Preprint

View full text Add to dashboard Cite

Point cloud registration is the basis for many robotic applications such as odometry and Simultaneous Localization And Mapping (SLAM), which are increasingly important for autonomous mobile robots. Computational resources and power budgets are limited on these robots, thereby motivating the development of resource-efficient registration method on low-cost FPGAs. In this paper, we propose a novel approach for FPGA-based 3D point cloud registration built upon a recent deep learning-based method, PointNetLK. A highly-efficient FPGA accelerator for PointNet-based feature extraction is designed and implemented on both low-cost and mid-range FPGAs (Avnet Ultra96v2 and Xilinx ZCU104). Our accelerator design is evaluated in terms of registration speed, accuracy, resource usage, and power consumption. Experimental results show that PointNetLK with our accelerator achieves up to 21.34x and 69.60x faster registration speed than the CPU counterpart and ICP, respectively, while only consuming 722mW and maintaining the same level of accuracy.

show abstract

An Optimized FPGA-Based Real-Time NDT for 3D-LiDAR Localization in Smart Vehicles

Cited by 11 publications

References 18 publications

Map representations for mobile robot localization

Map representations for mobile robot localization

A Universal LiDAR SLAM Accelerator System on Low-Cost FPGA

An Efficient Accelerator for Deep Learning-based Point Cloud Registration on FPGAs

Contact Info

Product

Resources

About