Scenario-based test is recognized as an effective way to evaluate and verify the Safety of the Intended Functionality (SOTIF) of an autonomous driving system, and a critical-scenario is beneficial to foreground the sy...
详细信息
Scenario-based test is recognized as an effective way to evaluate and verify the Safety of the Intended Functionality (SOTIF) of an autonomous driving system, and a critical-scenario is beneficial to foreground the system defect more obviously. However, for the state of the art, time consumption of critical-scenario-search in a real-time test is still high, especially for searching critical-scenarios with dissimilarities, which assists designers in discovering more failure modes of autonomous in limited time. To address above challenges, in this paper, we improve our previous scenario-searching approach considering dissimilarity assessments in iterations, we present a searching framework which combines control and exploration approaches, and implement the efficient evolution of scenarios with a backward target-tree search. We carry out a Vehicle-in-the-Loop (VIL) test with a real vehicle to verify our new approach, and the results show that the approach is able to search for scenarios directly against a black-box system in a real-time test, and achieve scenarios with both high criticality anddissimilarity, including unknown-unsafe cases. In our time-limited test, compared with different methods, our approach can achieve 1.9 times the amount of diversified scenarios with the almost same (slightly higher) criticality, either achieve 1.5 times the amount attached with 75% improvement in criticality. IEEE
In solving many-objective optimization problems(MaO Ps),existing nondominated sorting-based multi-objective evolutionary algorithms suffer from the fast loss of selection *** candidate solutions become nondominateddu...
详细信息
In solving many-objective optimization problems(MaO Ps),existing nondominated sorting-based multi-objective evolutionary algorithms suffer from the fast loss of selection *** candidate solutions become nondominatedduring the evolutionary process,thus leading to the failure of producing offspring toward Pareto-optimal front with *** we find a more effective way to select nondominated solutions andresolve this issue?To answer this critical question,this work proposes to evolve solutions through line complex rather than solution points in Euclidean ***,Plücker coordinates are used to project solution points to line complex composed of position vectors and momentum *** position vectors of the solution points,momentum vectors are used to extend the comparability of nondominated solutions and enhance selection ***,a new distance function designed for high-dimensional space is proposed to replace Euclidean distance as a more effective distancebased *** on them,a novel many-objective evolutionary algorithm(MaOEA)is proposed by integrating a line complex-based environmental selection strategy into the NSGAⅢ*** proposed algorithm is compared with the state of the art on widely used benchmark problems with up to 15 *** results demonstrate its superior competitiveness in solving MaOPs.
This study demonstrates and opens for academic research six sets of electric vehicle data collectedduring experiments on a low-adhesion road. The 11-dimensional data were collected from a small, single-motor SUV with...
详细信息
This study demonstrates and opens for academic research six sets of electric vehicle data collectedduring experiments on a low-adhesion road. The 11-dimensional data were collected from a small, single-motor SUV with front-wheel *** data collected for vehicles on low-adhesion roads are crucial forresearch on system dynamics modeling,stability control system design, and vehicle state estimation. Indeed, all vehicle control systems must be validated on low-adhesion roads, including vehicle models,
Few-shot semantic segmentation aims at training a model that can segment novel classes in a query image with only a few densely annotated support *** remains a challenge because of large intra-class variations between...
详细信息
Few-shot semantic segmentation aims at training a model that can segment novel classes in a query image with only a few densely annotated support *** remains a challenge because of large intra-class variations between the support and query *** approaches utilize 4d convolutions to mine semantic correspondence between the support and query ***,they still suffer from heavy computation,sparse correspondence,and large *** propose axial assembled correspondence network(AACNet)to alleviate these *** key point of AACNet is the proposed axial assembled 4d kernel,which constructs the basic block for semantic correspondence encoder(SCE).Furthermore,we propose the deblurring equations to provide more robust correspondence for the aforementioned SCE anddesign a novel fusion module to mix correspondences in a learnable *** on PASCAL-5~i reveal that our AACNet achieves a mean intersection-over-union score of 65.9%for 1-shot segmentation and 70.6%for 5-shot segmentation,surpassing the state-of-the-art method by 5.8%and 5.0%respectively.
This paper uses a nonlinear longitudinal control method fordistributeddrive electric vehicles(ddEV) to solve the safety problem of ddEV when driving on snow-covered ***,an optimal slip control scheme based on tire...
详细信息
ISBN:
(数字)9789887581536
ISBN:
(纸本)9781665482561
This paper uses a nonlinear longitudinal control method fordistributeddrive electric vehicles(ddEV) to solve the safety problem of ddEV when driving on snow-covered ***,an optimal slip control scheme based on tire force prediction is ***,use the Burckhardt tire model to draw μ-κ,curve,use the magic formula to draw F-κcurve,get the maximum longitudinal tire force that can be provided underdifferent road types *** calculate the tire force required for stable driving of the vehicle,predict the state of the vehicle in *** paper aims at the situation that tire force is not saturatedduring the driving process of the vehicle,the lower layer uses the nonlinear model predictive control(NMPC)algorithm to design the optimal slip *** main purpose is to control the torque of the four wheels simultaneously so that the slip rate can follow the optimal value,thereby effectively preventing the wheels from over-slipping during driving or locking during braking,enhancing the maneuverability and stability of the vehicle in extreme ***,the dynamic simulation software CarSim is used forsimulation verification under various working *** results show that the controllerdesigned in this paper can effectively track the optimal slip value of each wheel,play a driving assistance function,andrealize the multi-objective requirements of "safety-power-comfort".
diode rectifier unit(drU)-based high-voltage direct current(HVdC)transmissionsystems areeffectivein achieving the stableandeconomical operation of offshore *** theuncontrollable characteristicsof drUs,a grid-forming(G...
详细信息
diode rectifier unit(drU)-based high-voltage direct current(HVdC)transmissionsystems areeffectivein achieving the stableandeconomical operation of offshore *** theuncontrollable characteristicsof drUs,a grid-forming(GFM)strategy forwind-turbine converters isnecessary to support offshore AC voltageand ***,the active power-synchronization control in traditional GFM converters is unsuitable fordrU-based GFM ***,the stability issue fordrU-based HVdC systems involving drU-based GFM and grid-following(GFL)converters has not yet been *** solve these issues,this study begins with the characteristics of a drU-based HVdC system and presents a control scheme fordrU-based GFM converters for power ***,the dq-frame impedance model of the drU-based GFM converteris proposed for the stability analysis of the entire HVdC ***,a simulation platform is built to verify the model accuracy and system stability.
Infrareddetection technology has the advantages of all-weatherdetection and good concealment,which is widely used in long-distance target detection and tracking ***,the complex background,the strong noise,and the ch...
详细信息
Infrareddetection technology has the advantages of all-weatherdetection and good concealment,which is widely used in long-distance target detection and tracking ***,the complex background,the strong noise,and the characteristics of small scale and weak intensity of targets bring great difficulties to the detection of infrared small targets.A multi-channel based on attention network is proposed in this paper,aimed at the problem of high misseddetection rate and false alarm rate of traditional algorithms and the problem of large model,high complexity and poordetection performance of deep learning ***,given the difficulty in extracting the features of infrared multiscale and small dim targets,the multiple channels are designed based on dilated convolution to capture multiscale target ***,the coordinate attention block is incorporated in each channel to suppress background clutters adaptively and enhance target *** addition,the fusion of shallow detail features anddeep abstract semantic features is realized by synthesizing the contextual attention fusion ***,it is verified that,compared with other state-of-the-art methods based on the datasets SIrST and MdFA,the proposed algorithm further improves the detection effect,and the model size and computational complexity are smaller.
Low-frequency structure-borne noise in enclosed space is easy to produce booming sensation, which seriously affects the auditory comfort. But its sources are complex and the transmission path is strongly coupled, mean...
详细信息
The rapiddevelopment of autonomous driving technologies places higherdemands on the accurate estimation of tire-road friction coefficient (TrFC) in vehicle motion and vehicle dynamics control. A tire model must be b...
详细信息
In order to suppress the flux and torque ripples of direct torque control (dTC) for permanent magnet synchronous motors (PMSMs), a direct torque control based on voltage vector prediction is proposed in this paper. Fi...
详细信息
暂无评论