Point-Based Scan Matching by Differential Evolution

Krömer, Pavel; Konecny, Jaromír; Prauzek, Michal

doi:10.1109/incos.2016.62

Cited by 6 publications

(5 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section, the results of the computational experiments are detailed and the proposed evolutionary scan matching strategy based on cross-correlation (denoted CORR) is compared with a recent approach based on root mean square error (denoted RMSE) [44] and with the standard ICP algorithm [52]. Table 2 provides the results of the comparison between the ICP, RMSE and CORR scan matching strategies.…”

Section: Resultsmentioning

confidence: 99%

“…This work proposes a novel scan matching algorithm based on a combination of two recently introduced approaches. The cross-correlation approach [42,43] is adopted for the assessment of laser range scans' similarity and the differential evolution is used for the search of accurate laser range scan transformation parameters [44]. The evaluation of the proposed scan matching strategy is performed in a series of simulation experiments with the help of a software framework introduced in [45].…”

Section: Related Workmentioning

confidence: 99%

“…The definition of is essential for the properties of the process. A simple measure such as the root mean square error (RMSE) can provide a reasonable estimate of the degree of scan alignment and it can be used as a part of an optimization-based scan matching procedure [44]. However, its computational complexity is too high for practical use.…”

Section: Cross-correlationmentioning

confidence: 99%

“…In the proposed approach, the laser range scan is converted to Cartesian coordinates in the same manner as in [44]. Next, a grid map simplifying the scene is created from the point cloud data.…”

Section: Cross-correlationmentioning

confidence: 99%

“…A novel mobile robot self-localization strategy that uses the combination of cross-correlation of laser range scans [42] and the DE is designed and evaluated in this work. It improves a recent DE-based scan matching algorithm [44] through the use of laser range scan cross-correlation as an accurate and computationally lightweight fitness function. In this approach, the DE is used as an optimization mechanism to find the optimal set of parameters for an affine transformation that aligns two 2-dimensional laser range scans so that their mutual cross-correlation is maximized.…”

Section: Scan Matching By Cross-correlation and Differential Evolutionmentioning

confidence: 99%

See 4 more Smart Citations

Scan Matching by Cross-Correlation and Differential Evolution

et al. 2019

Self Cite

View full text Add to dashboard Cite

Scan matching is an important task, solved in the context of many high-level problems including pose estimation, indoor localization, simultaneous localization and mapping and others. Methods that are accurate and adaptive and at the same time computationally efficient are required to enable location-based services in autonomous mobile devices. Such devices usually have a wide range of high-resolution sensors but only a limited processing power and constrained energy supply. This work introduces a novel high-level scan matching strategy that uses a combination of two advanced algorithms recently used in this field: cross-correlation and differential evolution. The cross-correlation between two laser range scans is used as an efficient measure of scan alignment and the differential evolution algorithm is used to search for the parameters of a transformation that aligns the scans. The proposed method was experimentally validated and showed good ability to match laser range scans taken shortly after each other and an excellent ability to match laser range scans taken with longer time intervals between them.

show abstract

Section: Resultsmentioning

confidence: 99%