Gravel roads lack standardized features such as curbs or painted lines, presenting detection challenges to autonomous vehicles. Global Positioning Service (GPS) and high resolution maps may not be reliable for navigation of gravel roads, as some roads may only be width of the vehicle and GPS may not be accurate enough. Normal Distribution Transform (NDT) LiDAR scan matching may be insufficient for navigating on gravel roads as there may not be enough geometrically distinct features for reliable scan matching. This paper examined a method of classifying scanning LiDAR spatial and remission data features for explicit detection of unmarked gravel road surfaces. Exploration of terrain classification using high resolution scanning LiDAR data of specific road surfaces may allow for predicting gravel road boundary locations potentially enabling confident autonomous operations on gravel roads. The principal outcome of this work was a method for gravel road terrain detection using LiDAR data for the purpose of predicting potential road boundary locations. Random Decision Forests were trained using scanning LiDAR data terrain classification to detect unmarked gravel and asphalt surfaces. It was found that a true-positive accuracy for gravel and asphalt surfaces was 75% and 87% respectively at an estimated rate of 13 ms per 360 degree scan. Overlapping results between manually projected and actual road surface areas resulted in 93% intersecting gravel road detection accuracy. Automated post-process examination of classification results yielded an true-positive gravel road detection rate of 72%.
{"title":"Enhance Road Detection Data Processing of LiDAR Point Clouds to Specifically Identify Unmarked Gravel Rural Roads","authors":"Rhett Huston, Jay Wilhelm","doi":"10.1115/1.4066189","DOIUrl":"https://doi.org/10.1115/1.4066189","url":null,"abstract":"\u0000 Gravel roads lack standardized features such as curbs or painted lines, presenting detection challenges to autonomous vehicles. Global Positioning Service (GPS) and high resolution maps may not be reliable for navigation of gravel roads, as some roads may only be width of the vehicle and GPS may not be accurate enough. Normal Distribution Transform (NDT) LiDAR scan matching may be insufficient for navigating on gravel roads as there may not be enough geometrically distinct features for reliable scan matching. This paper examined a method of classifying scanning LiDAR spatial and remission data features for explicit detection of unmarked gravel road surfaces. Exploration of terrain classification using high resolution scanning LiDAR data of specific road surfaces may allow for predicting gravel road boundary locations potentially enabling confident autonomous operations on gravel roads. The principal outcome of this work was a method for gravel road terrain detection using LiDAR data for the purpose of predicting potential road boundary locations. Random Decision Forests were trained using scanning LiDAR data terrain classification to detect unmarked gravel and asphalt surfaces. It was found that a true-positive accuracy for gravel and asphalt surfaces was 75% and 87% respectively at an estimated rate of 13 ms per 360 degree scan. Overlapping results between manually projected and actual road surface areas resulted in 93% intersecting gravel road detection accuracy. Automated post-process examination of classification results yielded an true-positive gravel road detection rate of 72%.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"82 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141922533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the continuous development of various automobile technologies, the concept of intelligent automobile automatic driving has been introduced into people's lives, and it has great research value in traffic safety, traffic efficiency, and other aspects. Intelligent vehicles can accurately identify and track the target vehicle, which is one of the important preconditions for safe driving. However, a single tracking algorithm is often used in traditional intelligent vehicles with a low tracking accuracy under adverse circumstances. To solve this problem, a fusion tracking algorithm combining visual tracking and radar tracking algorithm is proposed, and intelligent vehicle target tracking technology is constructed based on the fusion algorithm. Through the performance comparison test, it was found that the fusion algorithm proposed in the study had the highest accuracy of 93% and the highest F measure of 0.98, both of which were superior to the comparison algorithm. Then, an empirical analysis is made of the target tracking technology proposed in the study. The results showed that the error range of yaw angle velocity of the target vehicle was −0.48 to 0.36, and the maximum root-mean-square error of lateral and longitudinal distance of the target vehicle detected by the technology was 0.03, which was superior to other tracking technologies. To sum up, the intelligent vehicle target tracking technology proposed in the research can improve the accuracy of intelligent vehicle target tracking and provide a guarantee for the safe driving of intelligent vehicles.
{"title":"Tracking Algorithm Application Integrating Visual and Radar Information in Intelligent Vehicle Target Tracking","authors":"Yu Wang, Jianfei Shi, Yu Zhao","doi":"10.1115/1.4066188","DOIUrl":"https://doi.org/10.1115/1.4066188","url":null,"abstract":"\u0000 With the continuous development of various automobile technologies, the concept of intelligent automobile automatic driving has been introduced into people's lives, and it has great research value in traffic safety, traffic efficiency, and other aspects. Intelligent vehicles can accurately identify and track the target vehicle, which is one of the important preconditions for safe driving. However, a single tracking algorithm is often used in traditional intelligent vehicles with a low tracking accuracy under adverse circumstances. To solve this problem, a fusion tracking algorithm combining visual tracking and radar tracking algorithm is proposed, and intelligent vehicle target tracking technology is constructed based on the fusion algorithm. Through the performance comparison test, it was found that the fusion algorithm proposed in the study had the highest accuracy of 93% and the highest F measure of 0.98, both of which were superior to the comparison algorithm. Then, an empirical analysis is made of the target tracking technology proposed in the study. The results showed that the error range of yaw angle velocity of the target vehicle was −0.48 to 0.36, and the maximum root-mean-square error of lateral and longitudinal distance of the target vehicle detected by the technology was 0.03, which was superior to other tracking technologies. To sum up, the intelligent vehicle target tracking technology proposed in the research can improve the accuracy of intelligent vehicle target tracking and provide a guarantee for the safe driving of intelligent vehicles.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"36 31","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141924751","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Taking the hybrid electric vehicle as the research object, under the premise of ensuring the braking safety, aiming at maximizing the use of motor regenerative braking force and improving the coordination performance of motor hydraulic braking, a simulation study of motor hydraulic braking control based on hybrid electric vehicle engine is proposed. According to the dynamic model and ideal braking force distribution curve of hybrid electric vehicle, combined with the common idea of electro-hydraulic compound braking force distribution, a three-layer braking control structure of hybrid electric vehicle is constructed. The management determines the braking intention through the driver's pedal action, calculates the expected torque, and the control layer obtains the target braking force distribution relationship through the logic gate limit control method based on the expected torque. According to the actual motor torque signal fed back by the executive layer and the wheel cylinder pressure signal of the hydraulic braking system, the braking force and regenerative braking force of the hydraulic system are dynamically coordinated and controlled to ensure that the state switching of each component can be rapid, stable and timely, and the control instruction is transmitted to the motor hydraulic braking system of the executive layer through the vehicle controller to complete the motor hydraulic braking of the hybrid electric vehicle engine. The experimental results show that this method can realize the reasonable distribution of motor hydraulic braking under different braking intensity, different initial braking speed and different pedal dip amplitude, which makes the reaction speed of hybrid electric vehicle in the braking process faster, the braking switching more stable and safe, effectively improves the energy utilization rate of hybrid electric vehicle, and ensures the economy and safety of braking control of hybrid electric vehicle.
{"title":"Simulation Study on Hydraulic Braking Control of Engine Motor of Hybrid Electric Vehicle","authors":"Fan Kang, Min Qiao","doi":"10.1115/1.4065936","DOIUrl":"https://doi.org/10.1115/1.4065936","url":null,"abstract":"\u0000 Taking the hybrid electric vehicle as the research object, under the premise of ensuring the braking safety, aiming at maximizing the use of motor regenerative braking force and improving the coordination performance of motor hydraulic braking, a simulation study of motor hydraulic braking control based on hybrid electric vehicle engine is proposed. According to the dynamic model and ideal braking force distribution curve of hybrid electric vehicle, combined with the common idea of electro-hydraulic compound braking force distribution, a three-layer braking control structure of hybrid electric vehicle is constructed. The management determines the braking intention through the driver's pedal action, calculates the expected torque, and the control layer obtains the target braking force distribution relationship through the logic gate limit control method based on the expected torque. According to the actual motor torque signal fed back by the executive layer and the wheel cylinder pressure signal of the hydraulic braking system, the braking force and regenerative braking force of the hydraulic system are dynamically coordinated and controlled to ensure that the state switching of each component can be rapid, stable and timely, and the control instruction is transmitted to the motor hydraulic braking system of the executive layer through the vehicle controller to complete the motor hydraulic braking of the hybrid electric vehicle engine. The experimental results show that this method can realize the reasonable distribution of motor hydraulic braking under different braking intensity, different initial braking speed and different pedal dip amplitude, which makes the reaction speed of hybrid electric vehicle in the braking process faster, the braking switching more stable and safe, effectively improves the energy utilization rate of hybrid electric vehicle, and ensures the economy and safety of braking control of hybrid electric vehicle.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"25 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141653623","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this study, we propose a robust and accurate SLAM method for dynamic environments. Our approach combines sparse optical flow with epipolar geometric constraints to detect motion, determining whether a priori dynamic objects are moving. By integrating semantic segmentation with this motion detection, we can effectively remove dynamic keypoints, eliminating the influence of dynamic objects. This dynamic object removal technique is integrated into ORB-SLAM2, en-hancing its robustness and accuracy for localization and mapping. Experimental results on the TUM dataset demonstrate that our proposed system significantly reduces pose estimation error compared to ORB-SLAM2. Specifically, the RMSE and standard deviation (S.D.) of ORB-SLAM2 are reduced by up to 97.78% and 97.91%, respectively, in highly dynamic se-quences, markedly improving robustness in dynamic environments. Furthermore, compared to other similar SLAM methods, our method reduces RMSE and S.D. by up to 69.26% and 73.03%, respectively. Dense semantic maps generated by our method also closely align with the ground truth.
在这项研究中，我们提出了一种针对动态环境的稳健而精确的 SLAM 方法。我们的方法结合了稀疏光流和外极几何约束来检测运动，先验地确定动态物体是否在移动。通过将语义分割与运动检测相结合，我们可以有效地去除动态关键点，从而消除动态物体的影响。这种动态物体移除技术被集成到 ORB-SLAM2 中，提高了定位和映射的鲁棒性和准确性。在TUM数据集上的实验结果表明，与ORB-SLAM2相比，我们提出的系统显著降低了姿势估计误差。具体来说，在高动态序列中，ORB-SLAM2 的 RMSE 和标准偏差（S.D.）分别降低了 97.78% 和 97.91%，显著提高了动态环境中的鲁棒性。此外，与其他类似的 SLAM 方法相比，我们的方法将 RMSE 和 S.D. 分别降低了 69.26% 和 73.03%。我们的方法生成的密集语义图也与地面实况非常吻合。
{"title":"Robust Visual SLAM in Dynamic Environment Based on Motion Detection and Segmentation","authors":"Xin Yu, Rulin Shen, Kang Wu, Zhi Lin","doi":"10.1115/1.4065873","DOIUrl":"https://doi.org/10.1115/1.4065873","url":null,"abstract":"\u0000 In this study, we propose a robust and accurate SLAM method for dynamic environments. Our approach combines sparse optical flow with epipolar geometric constraints to detect motion, determining whether a priori dynamic objects are moving. By integrating semantic segmentation with this motion detection, we can effectively remove dynamic keypoints, eliminating the influence of dynamic objects. This dynamic object removal technique is integrated into ORB-SLAM2, en-hancing its robustness and accuracy for localization and mapping. Experimental results on the TUM dataset demonstrate that our proposed system significantly reduces pose estimation error compared to ORB-SLAM2. Specifically, the RMSE and standard deviation (S.D.) of ORB-SLAM2 are reduced by up to 97.78% and 97.91%, respectively, in highly dynamic se-quences, markedly improving robustness in dynamic environments. Furthermore, compared to other similar SLAM methods, our method reduces RMSE and S.D. by up to 69.26% and 73.03%, respectively. Dense semantic maps generated by our method also closely align with the ground truth.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"63 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141688653","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pengfei Su, Natnael Getasew Tsehay, Wei Wang, Qixiang Zhao, Yangmin Li
Parking robots have been designed to alleviate parking stress in crowded areas. Existing models occupy large spaces and have limited adaptability to uneven ground. This paper aims to enhance the performance of parking robots by proposing a leader-follower-control based two-carrier Cooperative Parking Robot (CPR). In this system, two omnidirectional carriers operate on a tight cooperative transporting algorithm to achieve steady motion in their collaborative handling and transportation of the target car into the designated parking space. The novel CPR was designed, modeled, and implemented. The results indicate that the proposed CPR approached the targeted car and maintained a consistent position and heading angle in its cooperative parking operation at the speed of 0.6 m/s. The parking robot exhibited significant improvement in its adaptability to cars and uneven ground, and its compact configuration reduced its space occupation. Therefore, the proposed CPR has been proven robust for autonomous cooperative parking operations.
停车机器人的设计目的是缓解拥挤区域的停车压力。现有模型占用空间大，对不平地面的适应能力有限。本文提出了一种基于领导者-跟随者-控制的双载体合作停车机器人（CPR），旨在提高停车机器人的性能。在该系统中，两个全向载体采用紧密的合作运输算法，在合作搬运目标汽车并将其运送到指定停车位的过程中实现稳定运动。对新型 CPR 进行了设计、建模和实现。结果表明，拟议的 CPR 以 0.6 m/s 的速度接近目标汽车，并在协同泊车操作中保持一致的位置和方向角。该泊车机器人对汽车和不平地面的适应能力有了显著提高，其紧凑的结构也减少了对空间的占用。因此，所提出的 CPR 在自主合作停车操作中被证明是稳健的。
{"title":"Two-Carrier Cooperative Parking Robot: Design and Implementation","authors":"Pengfei Su, Natnael Getasew Tsehay, Wei Wang, Qixiang Zhao, Yangmin Li","doi":"10.1115/1.4065645","DOIUrl":"https://doi.org/10.1115/1.4065645","url":null,"abstract":"\u0000 Parking robots have been designed to alleviate parking stress in crowded areas. Existing models occupy large spaces and have limited adaptability to uneven ground. This paper aims to enhance the performance of parking robots by proposing a leader-follower-control based two-carrier Cooperative Parking Robot (CPR). In this system, two omnidirectional carriers operate on a tight cooperative transporting algorithm to achieve steady motion in their collaborative handling and transportation of the target car into the designated parking space. The novel CPR was designed, modeled, and implemented. The results indicate that the proposed CPR approached the targeted car and maintained a consistent position and heading angle in its cooperative parking operation at the speed of 0.6 m/s. The parking robot exhibited significant improvement in its adaptability to cars and uneven ground, and its compact configuration reduced its space occupation. Therefore, the proposed CPR has been proven robust for autonomous cooperative parking operations.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"28 11","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141273047","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Christopher T. Goodin, Lucas Cagle, Greg Henley, Brandon Black, Justin Carrillo, David P. McInnis
This paper presents a study of how communication ranges influence the performance of a new decentralized control method for swarms of autonomously navigating ground vehicles that uses a blended leader-follower / artificial potential field approach. While teams of autonomous ground vehicles (AGV) that can navigate autonomously through off-road terrain have a variety of potential uses, it may be difficult to control the team in low-infrastructure environments that lack long-range radio communications capabilities. In this work, we propose a novel decentralized swarm control algorithm that combines the potential-field planning method with the leader-follower control algorithm and biologically-inspired inter-robot interactions to effectively control the navigation of a team of AGV (swarm) through rough terrain using only a single lead vehicle. We use simulated experimentation to demonstrate the robustness of this approach using only point-to-point wireless communication with realistic communication ranges. Furthermore, we analyze the range requirements of the communication network as the number in the swarm increases. We find that wireless communication range must increase as the number of agents in the swarm increases in order to effectively control the swarm. Our analysis showed that mission success decreased by 40% when the communication range was reduced from 100 meters to 200 meters, with the exact reduction also depending on the number of vehicles.
{"title":"Decentralized Swarm Control in Communication-Constrained Environments Using a Blended Leader Follower-Artificial Potential Field with Biologically Inspired Interactions","authors":"Christopher T. Goodin, Lucas Cagle, Greg Henley, Brandon Black, Justin Carrillo, David P. McInnis","doi":"10.1115/1.4065533","DOIUrl":"https://doi.org/10.1115/1.4065533","url":null,"abstract":"\u0000 This paper presents a study of how communication ranges influence the performance of a new decentralized control method for swarms of autonomously navigating ground vehicles that uses a blended leader-follower / artificial potential field approach. While teams of autonomous ground vehicles (AGV) that can navigate autonomously through off-road terrain have a variety of potential uses, it may be difficult to control the team in low-infrastructure environments that lack long-range radio communications capabilities. In this work, we propose a novel decentralized swarm control algorithm that combines the potential-field planning method with the leader-follower control algorithm and biologically-inspired inter-robot interactions to effectively control the navigation of a team of AGV (swarm) through rough terrain using only a single lead vehicle. We use simulated experimentation to demonstrate the robustness of this approach using only point-to-point wireless communication with realistic communication ranges. Furthermore, we analyze the range requirements of the communication network as the number in the swarm increases. We find that wireless communication range must increase as the number of agents in the swarm increases in order to effectively control the swarm. Our analysis showed that mission success decreased by 40% when the communication range was reduced from 100 meters to 200 meters, with the exact reduction also depending on the number of vehicles.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"15 11","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141119969","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This article represents an extensive literature on tire hydroplaning, specifically focusing on the assessment of real-time estimation methodologies and numerical modeling for both partial and total hydroplaning phenomenon. Hydroplaning still poses a significant challenge for contemporary passenger cars, even those equipped with state of the art safety systems. The active safety features that equip the most technologically advanced passenger cars are unable to forecast and prevent the occurrence of hydroplaning. Total hydroplaning represents a phenomenon which occurs when the tire reaches a point where it can no longer expel the water from its tread grooves, leading to a complete control loss of the motor vehicle. This describes a scenario in which the entire contact patch is lifted from the ground due to the hydrodynamic forces generated at the contact between the tire and the layer of water formed on the road. Nevertheless, the decrease in contact between the tire and the road surface occurs gradually, a phenomenon which is presented in literature as partial hydroplaning. The longitudinal speed that marks the transition from partial hydroplaning to total hydroplaning is defined as the critical hydroplaning speed. These principles are widely acknowledged among researchers in the hydroplaning field. Nonetheless, the literature review reveals variations for defining the critical hydroplaning speed threshold across different experimental investigations. In this article, past studies, and state-of-the-art research on tire hydroplaning has been reviewed, especially focusing on real-time estimation methodologies and numerical modeling of the partial and of the total hydroplaning phenomenon.
{"title":"HYDROPLANING OF TIRES: A REVIEW OF NUMERICAL MODELING AND NOVEL SENSING METHODS","authors":"Alexandru Vilsan, Corina Sandu","doi":"10.1115/1.4065379","DOIUrl":"https://doi.org/10.1115/1.4065379","url":null,"abstract":"\u0000 This article represents an extensive literature on tire hydroplaning, specifically focusing on the assessment of real-time estimation methodologies and numerical modeling for both partial and total hydroplaning phenomenon. Hydroplaning still poses a significant challenge for contemporary passenger cars, even those equipped with state of the art safety systems. The active safety features that equip the most technologically advanced passenger cars are unable to forecast and prevent the occurrence of hydroplaning. Total hydroplaning represents a phenomenon which occurs when the tire reaches a point where it can no longer expel the water from its tread grooves, leading to a complete control loss of the motor vehicle. This describes a scenario in which the entire contact patch is lifted from the ground due to the hydrodynamic forces generated at the contact between the tire and the layer of water formed on the road. Nevertheless, the decrease in contact between the tire and the road surface occurs gradually, a phenomenon which is presented in literature as partial hydroplaning. The longitudinal speed that marks the transition from partial hydroplaning to total hydroplaning is defined as the critical hydroplaning speed. These principles are widely acknowledged among researchers in the hydroplaning field. Nonetheless, the literature review reveals variations for defining the critical hydroplaning speed threshold across different experimental investigations. In this article, past studies, and state-of-the-art research on tire hydroplaning has been reviewed, especially focusing on real-time estimation methodologies and numerical modeling of the partial and of the total hydroplaning phenomenon.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"34 9","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140672039","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The collision damage of automated cars has grown in importance as self-driving car technology has advanced to the pilot operation stage. The study builds a collision damage prediction model for automated driving cars, optimized deep convolutional neural networks using the self-attention mechanism, and designs a degree convolutional neural network algorithm incorporating the attention mechanism in order to avoid the dangers that will be encountered on the way to automated driving in advance. The findings demonstrated that the four index values of the modified algorithm in the calculation of the index were, respectively, 94.0%, 94.8%, 93.6%, and 0.88, with higher overall performance. The prediction model's accuracy during training on the training data set and validation data set was 100% and 98%, respectively, demonstrating its efficacy. The prediction model's prediction accuracy in calculating the degree of auto collision damage for 10 working conditions in the validation dataset is 83.3%, and the prediction results are essentially consistent with the trend of the actual collision damage degree curve, demonstrating both the viability and high prediction accuracy of the prediction model. The aforementioned findings demonstrated the model's strong performance and great application value in the field of self-driving car collision avoidance and warning.
{"title":"Design and Application of Deep Learning-based Crash Damage Prediction Model for Self-Driving Cars","authors":"Wenxia Zhang, Zhixue Wang","doi":"10.1115/1.4065307","DOIUrl":"https://doi.org/10.1115/1.4065307","url":null,"abstract":"\u0000 The collision damage of automated cars has grown in importance as self-driving car technology has advanced to the pilot operation stage. The study builds a collision damage prediction model for automated driving cars, optimized deep convolutional neural networks using the self-attention mechanism, and designs a degree convolutional neural network algorithm incorporating the attention mechanism in order to avoid the dangers that will be encountered on the way to automated driving in advance. The findings demonstrated that the four index values of the modified algorithm in the calculation of the index were, respectively, 94.0%, 94.8%, 93.6%, and 0.88, with higher overall performance. The prediction model's accuracy during training on the training data set and validation data set was 100% and 98%, respectively, demonstrating its efficacy. The prediction model's prediction accuracy in calculating the degree of auto collision damage for 10 working conditions in the validation dataset is 83.3%, and the prediction results are essentially consistent with the trend of the actual collision damage degree curve, demonstrating both the viability and high prediction accuracy of the prediction model. The aforementioned findings demonstrated the model's strong performance and great application value in the field of self-driving car collision avoidance and warning.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"55 7","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140709787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Thomas Most, Maximillian Rasch, Paul Tobe Ubben, Roland Niemeier, Veit Bayer
In this paper, we present a stochastic approach for the reliability evaluation of specific traffic scenarios as one component in the validation procedure of Advanced Driver Assistance Systems (ADAS). In this analysis, the control device is represented as a simulation model using software-in-the-loop technology. Specific inputs of this simulated controller are modeled as scalar random inputs. Based on a definition of a failure criterion, the well known reliability method can be applied. In the present paper, a variance reduced importance sampling strategy for multiple failure regions is presented, which was developed for a scenario-based safety assessment framework.
{"title":"A multi-modal importance sampling approach for the probabilistic safety assessment of automated driver assistance systems","authors":"Thomas Most, Maximillian Rasch, Paul Tobe Ubben, Roland Niemeier, Veit Bayer","doi":"10.1115/1.4065308","DOIUrl":"https://doi.org/10.1115/1.4065308","url":null,"abstract":"\u0000 In this paper, we present a stochastic approach for the reliability evaluation of specific traffic scenarios as one component in the validation procedure of Advanced Driver Assistance Systems (ADAS). In this analysis, the control device is represented as a simulation model using software-in-the-loop technology. Specific inputs of this simulated controller are modeled as scalar random inputs. Based on a definition of a failure criterion, the well known reliability method can be applied. In the present paper, a variance reduced importance sampling strategy for multiple failure regions is presented, which was developed for a scenario-based safety assessment framework.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"20 19","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140711819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Achieving human-level driving performance in complex environments remains a major challenge in the field of Deep Learning (DL) based end-to-end Autonomous Driving Systems (ADS). In ADS, generalization to rare edge cases poses a serious safety concern with DL based models. The leading solution to this problem is scaling; the construction of larger models and datasets. However, limitations in the computational power available to autonomous vehicles, coupled with the under-representation of safety-critical edge cases in large autonomous driving datasets raise questions over the suitability of scaling for ADS. In this work, we investigate the performance of an alternate, computationally less demanding, Machine Learning (ML) algorithm, Hierarchical Temporal Memory (HTM). Existing HTM models use rudimentary encoding schemes that have thus far limited their application to simple inputs. Motivated by this shortcoming, we first propose a bespoke CNN based encoding scheme suited to the input data used in ADS. We then integrate this encoding scheme into a novel DL-HTM end-to-end ADS. The proposed DL-HTM based end-to-end ADS is trained and evaluated against a conventional DL end-to-end ADS based on the widely used AlexNet model from the literature. Our evaluation results show that the proposed DL-HTM model achieves comparable performance with far fewer trainable parameters than the conventional DL based end-to-end ADS. Results also indicate that the proposed model demonstrates a superior capacity for learning under-represented classes, i.e. edge cases, in the dataset.
{"title":"A Hierarchical Temporal Memory Based End-to-End Autonomous Driving System","authors":"Luc Le Mero, M. Dianati, Graham Lee","doi":"10.1115/1.4064989","DOIUrl":"https://doi.org/10.1115/1.4064989","url":null,"abstract":"\u0000 Achieving human-level driving performance in complex environments remains a major challenge in the field of Deep Learning (DL) based end-to-end Autonomous Driving Systems (ADS). In ADS, generalization to rare edge cases poses a serious safety concern with DL based models. The leading solution to this problem is scaling; the construction of larger models and datasets. However, limitations in the computational power available to autonomous vehicles, coupled with the under-representation of safety-critical edge cases in large autonomous driving datasets raise questions over the suitability of scaling for ADS. In this work, we investigate the performance of an alternate, computationally less demanding, Machine Learning (ML) algorithm, Hierarchical Temporal Memory (HTM). Existing HTM models use rudimentary encoding schemes that have thus far limited their application to simple inputs. Motivated by this shortcoming, we first propose a bespoke CNN based encoding scheme suited to the input data used in ADS. We then integrate this encoding scheme into a novel DL-HTM end-to-end ADS. The proposed DL-HTM based end-to-end ADS is trained and evaluated against a conventional DL end-to-end ADS based on the widely used AlexNet model from the literature. Our evaluation results show that the proposed DL-HTM model achieves comparable performance with far fewer trainable parameters than the conventional DL based end-to-end ADS. Results also indicate that the proposed model demonstrates a superior capacity for learning under-represented classes, i.e. edge cases, in the dataset.","PeriodicalId":164923,"journal":{"name":"Journal of Autonomous Vehicles and Systems","volume":"18 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140398230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：