Model-based techniques have proven to be successful in interpreting the large amount of information contained in images. Associated fitting algorithms search for the global optimum of an objective function, which should correspond to the best model fit in a given image. Although fitting algorithms have been the subject of intensive research and evaluation, the objective function is usually designed ad hoc, based on implicit and domain-dependent knowledge. In this article, we address the root of the problem by learning more robust objective functions. First, we formulate a set of desirable properties for objective functions and give a concrete example function that has these properties. Then, we propose a novel approach that learns an objective function from training data generated by manual image annotations and this ideal objective function. In this approach, critical decisions such as feature selection are automated, and the remaining manual steps hardly require domain-dependent knowledge. Furthermore, an extensive empirical evaluation demonstrates that the obtained objective functions yield more robustness. Learned objective functions enable fitting algorithms to determine the best model fit more accurately than with designed objective functions.
Abstract-In this paper, we present a vehicle safety application based on data gathered by a laser scanner and two short range radars that recognizes unavoidable collisions with stationary objects before they take place in order to trigger restraint systems. Two different software modules are compared that perform the processing of raw data and deliver a description of the vehicle's environment. A comprehensive experimental evaluation based on relevant crash and non-crash scenarios is presented.
Objective: Left turn across path with traffic from the opposite direction (LTAP/OD) is the second most frequent car-to-car intersection crash type after straight crossing path (SCP) in Germany and the United States. Intersection automated emergency braking (AEB) for passenger cars can address these crashes. This study investigates 2 implementation strategies of intersection AEB addressing LTAP/OD crashes: (1) only the turning car is equipped with an intersection AEB and (2) turning and straightheading cars are equipped with an intersection AEB. For each strategy, the influence of a safety zone around the vehicles that should not be entered is evaluated in terms of accident avoidance, injury mitigation, and change in velocity (delta-V) of remaining accidents. Results are given as a function of market penetration. Methods: A total of 372 LTAP/OD crashes from the time series precrash matrix (PCM), a subsample of the German In-Depth Accident Study (GIDAS), were resimulated in the PRediction of Accident Evolution by Diversification of Influence factors in COmputer simulation (PRAEDICO) simulation framework. A Kudlich-Slibar rigid-body impact model and an injury risk curve derived from GIDAS were used to predict remaining moderate to fatal (Maximum Abbreviated Injury Scale [MAIS] 2 þ F) injuries among car occupants. Results: With a safety zone of 0.2 m, when the turning vehicle only was equipped with an intersection AEB, 59% of the crashes were avoided at a 100% market penetration. With both vehicles equipped the percentage increased to 77%. MAIS 2 þ F injured occupants were reduced by 60 and 76%, respectively. Considering both the turning and the straight-heading vehicles, the delta-V decreased strongly with market penetration in remaining left-side impacts but only slightly in remaining frontal and right-side impacts. Eliminating the safety zone substantially decreases effectiveness in all conditions. Conclusions: Implementation strategy and safety zone definition strongly influence the real-life performance of intersection AEB. AEB should be applied not only for the turning vehicle but also for the straight-going vehicle to benefit from the full potential. Situationally appropriate safety zone definitions, in line with human hazard perception, need more attention and are a key to balance true positive and false positive performance. Remaining delta-V does not decrease broadly; hence, there is no evidence that future LTAP/OD crashes will be generally of lower severity. This highlights the need for continuous development of in-crash protection.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.