浏览全部资源
扫码关注微信
纸质出版日期： 2021-01-16 ，
录用日期： 2020-10-26
移动端阅览
崔明月, 钟仕鹏, 刘思瑶, 李博洋, 吴成昊, 黄凯. 利用边缘计算的多车协同激光雷达SLAM[J]. 中国图象图形学报, 2021,26(1):218-228.
Mingyue Cui, Shipeng Zhong, Siyao Liu, Boyang Li, Chenghao Wu, Kai Huang. Cooperative LiDAR SLAM for multi-vehicles based on edge computing[J]. Journal of Image and Graphics, 2021,26(1):218-228.
目的
2
激光雷达实时定位与建图（simultaneous localization and mapping，SLAM）是智能机器人领域的重要组成部分，通过对周边环境的3维建模，可以实现无人驾驶车辆的自主定位和精准导航。针对目前单个车辆激光雷达建图周期长、算力需求大的现状，提出了基于边缘计算的多车协同建图方法，能够有效地负载均衡，在保证单个车辆精准定位的同时，增加多个车辆之间的地图重用性。
方法
2
构建基于阈值的卸载函数，论证边缘计算下的多车卸载决策属于势博弈问题，设计实现基于边缘计算的势博弈卸载算法，在模型具有纳什均衡的基础上实现任务调度，引入
$$\alpha $$
-Nash最佳响应动态加速算法收敛，并采用由粗到细的点云匹配方法提高地图匹配性能，实现车辆的精准定位。最后，基于地图的相对可信度，高效地合并基站覆盖范围内的多个车辆的建图数据。
结果
2
实验表明，基于博弈论的调度方法在保证定位可靠性的前提下，能够有效地实现多车协同SLAM，且多车协同的定位与建图结果与使用载波相位差分技术（real-time kinematic，RTK）的高精度差分全球定位系统（differential global positioning system，DGPS）结果足够接近，相比于单车建图而言，横向定位和纵向定位的平均精度分别提高了6.0倍和3.9倍。
结论
2
本文方法解决了基于边缘计算的多车协同激光雷达SLAM问题，借助边缘服务器的计算资源，无人驾驶车辆可以有效地减少本地资源需求和定位延迟。该方法通过各个车辆之间的资源博弈，最终实现纳什均衡。实现基于边缘计算的激光雷达定位服务，且高效地完成多车之间的地图合并，仿真和真实环境中的实验表明了方法的有效性。
Objective
2
LiDAR simultaneous localization and mapping (SLAM) is an important component of the field of intelligent robotics. The robot should be able to perceive the information of the surrounding environment and accurately locate its own position
which is also the premise for the robot to autonomously navigate. Building the entire map requires the single vehicle to drive all over the whole area
which makes the mapping periods longer. In addition
more data require more computing power for onboard units. To solve the above problems
a cooperative LiDAR SLAM for multi-vehicles based on edge computing method is proposed. This method can make the load balance by offloading tasks to the edge server. Aside from ensuring accurate localization of the single vehicle
it can also increase the reusability of the mapping results for multi-vehicles. This study models the computation offloading decision-making problem among multi-vehicles as a task offloading game. It designs the offloading algorithm based on the potential game to compute the task scheduling sequence. The concept of relative confidence is introduced to reduce the odometer error in the process of mapping as much as possible
which also makes our merged maps of multi-vehicles more accurate.
Method
2
First
this study constructs a threshold-based offloading function that includes latency constraint and signal quality constraint. Then
this study proves that the minimum latency problem of multi-vehicles is a potential game. The potential game always reaches the Nash equilibrium and has limited improved properties. Therefore
the relevant strategy based on the potential game is designed for offloading tasks. With the characteristics of the Nash equilibrium
our algorithm ensures that vehicles in equilibrium can obtain mutually satisfactory solutions. This study introduces the concept of
$$\alpha $$
-Nash dynamics to speed up the convergence of the algorithm. Then
a coarse-to-fine point cloud matching scheme is used to realize the map matching. During the coarse matching phase
the point cloud data of different vehicles are initially matched. This study uses the fast point feature histogram algorithm to extract key points and descriptors from the point cloud and the distance between the point pair within the spherical neighborhood of the key point. Then
the random sample consensus(RANSAC) method is used to estimate the point-to-point correspondences between the key points
which can obtain the initial matrix rotation and translation matrix that describes the rough relative transformation of two maps
. In the fine matching phase
the iterative closest-point algorithm is used to further optimize the matching results and increase accuracy. On the basis of this method
matching two-frame point clouds can converge faster and achieve a better matching result. Finally
local LiDAR maps are merged based on relative confidence. The node with the lowest weight is selected
then the node set connected with the node is sorted according to the relative confidence and the size of the overlapping area. Finally
the map is merged according to the sorted order
and the above steps are repeated until all nodes are traversed.
Result
2
To verify the effectiveness of this study
we conducted simulation and real-world experiments. In the simulation experiment
$$\alpha $$
is set to different values
and experiments are conducted with different numbers of vehicles. This study takes the average number of time slots and the average system-wide overhead as two important performance metrics
which indicate the convergence speed of the algorithm and the efficiency of reducing system overhead
respectively. Results show that as
$$\alpha $$
decreases
the convergence speed increases
while the efficiency of reducing the system overhead decreases. To balance the two performance metrics mentioned above
the system sets the parameter
$$\alpha $$
to 0.8. Results also show that under the same parameter
the average number of time slots and the average system-wide overhead increase almost linearly as the number of vehicles increases. This finding shows that our algorithm adapts well with different sizes of users. In the real-world scene
a 1.49 km campus road is chosen as the test section
including turning
going straight
and changing lanes. Three base stations and four autonomous vehicles with identical hardware configurations are depl
oyed in the test environment. The test area is divided into three parts with the base station as the center because of the limitation of the actual number of vehicles. Three to four vehicles are presented in each area to simulate the effect of mapping results for nine vehicles. The experiment result shows that the trajectory of the cooperative mapping are closely matched with that of real-time kinematic
while the deviation for single-vehicle mapping is much larger. In addition
compared with single-vehicle mapping
the average accuracy of latitude and longitude localization for cooperative mapping is improved about 6.0 and 3.9 times
respectively. In the worst case
the accuracy improves about 1.7 times in latitude and 1.6 times in longitude.
Conclusion
2
This paper proposes an approach to offloading SLAM tasks based on edge computing in the multi-vehicle scenario. On the basis of edge computing
autonomous vehicles can effectively reduce local computing power and the latency of localization. This method achieves Nash equilibrium by means of resource games among several vehicles. While each vehicle realizes its accurate localization
it collaborates with other vehicles to complete the overall map construction. Simulation and experimental deployment in the real environment demonstrate the effectiveness of our method. We believe that the cooperative LiDAR SLAM for multi-vehicles based on edge computing can effectively increase the quality of service for autonomous driving.
边缘计算激光雷达实时定位与构图(SLAM)任务卸载多车协同无人驾驶
edge computingsimultaneous localization and mapping(SLAM)task offloadingmulti-vehicle collaborationautonomous driving
Awerbuch B, Azar Y, Epstein A, Mirrokni V S and Skopalik A. 2008. Fast convergence to nearly optimal solutions in potential games//Proceedings of the 9th ACM Conference on Electronic Commerce. Chicago, USA: ACM: 264-273[DOI:10.1145/1386790.1386832http://dx.doi.org/10.1145/1386790.1386832]
Biber P and Strasser W. 2003. The normal distributions transform: a new approach to laser scan matching//Proceedings of 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems. Las Vegas, USA: IEEE: 2743-2748[DOI:10.1109/IROS.2003.1249285http://dx.doi.org/10.1109/IROS.2003.1249285]
Chen X, Jiao L, Li W Z and Fu X M. 2016. Efficient multi-user computation offloading for mobile-edge cloud computing. IEEE/ACM Transactions on Networking, 24(5):2795-2808[DOI:10.1109/TNET.2015.2487344]
Chen X F, Zhang H G, Wu C, Mao S W, Ji Y S and Bennis M. 2019. Optimized computation offloading performance in virtual edge computing systems via deep reinforcement learning. IEEE Internet of Things Journal, 6(3):4005-4018[DOI:10.1109/JIOT.2018.2876279]
Cui Y, Song J, Ren K, Li M M, Li Z P, Ren Q M and Zhang Y J. 2017. Software defined cooperative offloading for mobile cloudlets. IEEE/ACM Transactions on Networking, 25(3):1746-1760[DOI:10.1109/TNET.2017.2650964]
Deng S G, Huang L T, Taheri J and Zomaya A Y. 2015. Computation offloading for service workflow in mobile cloud computing. IEEE Transactions on Parallel and Distributed Systems, 26(12):3317-3329[DOI:10.1109/TPDS.2014.2381640]
Holz D, Ichim A E, Tombari F, Rusu R B and Behnke S. 2015. Registration with the point cloud library:a modular framework for aligning in 3-D. IEEE Robotics and Automation Magazine, 22(4):110-124[DOI:10.1109/MRA.2015.2432331]
Li X L, Yang B W, Xie X H, Li D and Xu L J. 2018. Influence of waveform characteristics on LiDAR ranging accuracy and precision. Sensors, 18(4):#1156[DOI:10.3390/s18041156]
Liu Y, Yu H M, Xie S L and Zhang Y. 2019. Deep reinforcement learning for offloading and resource allocation in vehicle edge computing and networks. IEEE Transactions on Vehicular Technology, 68(11):11158-11168[DOI:10.1109/TVT.2019.2935450]
Min M H, Xiao L, Chen Y, Cheng P, Wu D and Zhuang W H. 2019. Learning-based computation offloading for IoT devices with energy harvesting. IEEE Transactions on Vehicular Technology, 68(2):1930-1941[DOI:10.1109/TVT.2018.2890685]
Monderer D and Shapley L S. 1996. Potential games. Games and Economic Behavior, 14(1):124-143[DOI:10.1006/game.1996.0044]
Moubayed A, Shami A, Heidari P, Larabi A and Brunner R. 2020. Edge-enabled V2X service placement for intelligent transportation systems[EB/OL].[2020-08-15].https://arxiv.org/pdf/2001.06288.pdfhttps://arxiv.org/pdf/2001.06288.pdf
Qi Q, Wang J Y, Ma Z Y, Sun H F, Cao Y F, Zhang L X and Liao J X. 2019. Knowledge-driven service offloading decision for vehicular edge computing:a deep reinforcement learning approach. IEEE Transactions on Vehicular Technology, 68(5):4192-4203[DOI:10.1109/TVT.2019.2894437]
Quek T Q S, de la Roche G, Güvenç I and Kountouris M. 2013. Small Cell Networks:Deployment, PHY Techniques, and Resource Management. Cambridge:Cambridge University Press:260-279
Rodrigues T G, Suto K, Nishiyama H, Kato N and Temma K. 2018. Cloudlets activation scheme for scalable mobile edge computing with transmission power control and virtual machine migration. IEEE Transactions on Computers, 67(9):1287-1300[DOI:10.1109/TC.2018.2818144]
Rusinkiewicz S and Levoy M. 2002. Efficient variants of the ICP algorithm//Proceedings of the 3rd International Conference on 3-D Digital Imaging and Modeling. Quebec, Canada: IEEE: 145-152[DOI:10.1109/IM.2001.924423http://dx.doi.org/10.1109/IM.2001.924423]
Rusu R B, Blodow N and Beetz M. 2009. Fast point feature histograms (FPFH) for 3D registration//Proceedings of 2009 IEEE International Conference on Robotics and Automation. Kobe, Japan: IEEE: 3212-3217[DOI:10.1109/ROBOT.2009.5152473http://dx.doi.org/10.1109/ROBOT.2009.5152473]
Wang Y P, Lang P, Tian D X, Zhou J S, Duan X T, Cao Y and Zhao D Z. 2020. A game-based computation offloading method in vehicular multiaccess edge computing networks. IEEE Internet of Things Journal, 7(6):4987-4996[DOI:10.1109/JIOT.2020.2972061]
Wei Z L, Zhao B K, Su J S and Lu X C. 2019. Dynamic edge computation offloading for internet of things with energy harvesting:a learning method. IEEE Internet of Things Journal, 6(3):4436-4447[DOI:10.1109/JIOT.2018.2882783]
Yang L, Cao J N, Wang Z Y andWu W G. 2017. Network aware multi-user computation partitioning in mobile edge clouds//Proceedings of the 46th International Conference on Parallel Processing. Bristol, UK: IEEE: 302-311[DOI:10.1109/ICPP.2017.39http://dx.doi.org/10.1109/ICPP.2017.39]
Yang L C, Zhang H L, Li X, Ji H and Leung V C M. 2018. A distributed computation offloading strategy in small-cell networks integrated with mobile edge computing. IEEE/ACM Transactions on Networking, 26(6):2762-2773[DOI:10.1109/TNET.2018.2876941]
Zhang J and Singh S. 2014. LOAM: Lidar odometry and mapping in real-time//Proceedings of Robotics: Science and Systems X. Berkeley, USA: IEEE: 109-111[DOI:10.15607/RSS.2014.X.00http://dx.doi.org/10.15607/RSS.2014.X.00]
Zhang K, Zhu Y X, Leng S P, He Y J, Maharjan S and Zhang Y. 2019. Deep learning empowered task offloading for mobile edge computing in urban informatics. IEEE Internet of Things Journal, 6(5):7635-7647[DOI:10.1109/JIOT.2019.2903191]
Zhe X Y, Li B Y, Zhang X Y, Chen L and Huang K. 2018. Online cooperative 3D mapping for autonomous driving//2018 IEEE Intelligent Vehicles Symposium. Changshu, China: IEEE: 256-261[DOI:10.1109/IVS.2018.8500615http://dx.doi.org/10.1109/IVS.2018.8500615]
相关文章
相关作者
相关机构
京公网安备11010802024621