Realistic urban scene generation has been extensively studied for the sake of the development of autonomous vehicles. However, the research has primarily focused on the synthesis of vehicles and pedestrians, while the...Realistic urban scene generation has been extensively studied for the sake of the development of autonomous vehicles. However, the research has primarily focused on the synthesis of vehicles and pedestrians, while the generation of cyclists is rarely presented due to its complexity. This paper proposes a perspective-aware and realistic cyclist generation method via object retrieval. Images, semantic maps, and depth labels of objects are first collected from existing datasets, categorized by class and perspective, and calculated by an algorithm newly designed according to imaging principles. During scene generation, objects with the desired class and perspective are retrieved from the collection and inserted into the background, which is then sent to the modified 2D synthesis model to generate images. This pipeline introduces a perspective computing method, utilizes object retrieval to control the perspective accurately, and modifies a diffusion model to achieve high fidelity. Experiments show that our proposal gets a 2.36 Fréchet Inception Distance, which is lower than the competitive methods, indicating a superior realistic expression ability. When these images are used for augmentation in the semantic segmentation task, the performance of ResNet-50 on the target class can be improved by 4.47%. These results demonstrate that the proposed method can be used to generate cyclists in corner cases to augment model training data, further enhancing the perception capability of autonomous vehicles and improving the safety performance of autonomous driving technology.展开更多
In order to analyze the risky factors that affect vehicle-cyclist crash injury severity at the intersection area,especially the factors relating to the road users behaviors,an empirical study was conducted by collecti...In order to analyze the risky factors that affect vehicle-cyclist crash injury severity at the intersection area,especially the factors relating to the road users behaviors,an empirical study was conducted by collecting accident records from 2011 to 2015 from the General Estimates System.After preliminary screening,the variables were classified into 5 main categories including cyclists characteristic and behavior,drivers characteristic and behavior,vehicle characteristic,intersection condition,and time.The random parameter ordinal probit(RPOP)was used to study the significant influencing factors and corresponding heterogeneity.The results show that failing to obey traffic signals,failing to yield to right-of-way,dash and drinking before cycling can increase the injury severity for cyclists,and the corresponding fatal injury likelihoods increase by 53.2%,40.0%,86.3%,and 211.5%,respectively.Moreover,drivers inattention,speeding,going straight and left turning increase the risk of crashing for cyclists.The corresponding fatal injury likelihoods increase by 134.5%,186.5%,69.3%,and 22.7%,respectively.Other indicators such as age,gender,vehicle type,traffic signal and intersection type can also affect injury severity.展开更多
This work aimed to explore the switch tendency of bicycle use between cyclists and non-cyclists.The attitude based market segmentation approach was proposed to achieve the research objective.The filed investigations w...This work aimed to explore the switch tendency of bicycle use between cyclists and non-cyclists.The attitude based market segmentation approach was proposed to achieve the research objective.The filed investigations were conducted in Nanjing,China,to obtain travelers' actions and attitudes towards bicycle uses.The structural equation modeling(SEM) was used to identify the attitudinal factors indicating variables and to explore the interrelationships among them.The SEMs were developed separately for the cyclist group and the non-cyclist group.All respondents were clustered into eight distinct segments by six selected attitudinal factors.The mode switch tendency and attitude in each segment is different from others,indicating that different segments have particular policies or strategies to encourage cycling.Policy implications that best serve the potential bicycle users were discussed to reduce the number of cyclists who have high tendency to use other modes,and increase the possibility of using bicycle in the non-cyclists group with the moderate and high switch tendency.展开更多
The concept of facial impact protection mask for cyclists is proposed in response to increased participation in cycling and the need for injury prevention. The research aims to develop an approach for design of facial...The concept of facial impact protection mask for cyclists is proposed in response to increased participation in cycling and the need for injury prevention. The research aims to develop an approach for design of facial impact protection gear to reduce the risk of severe injury. Impact test equipment and procedure, face surrogate and protection material performance criteria are developed. Three groups of protective materials – rigid crushable, semi rigid, and soft cushion foams are tested and assessed according to criteria. The criteria are linked to measures of the risk of facial and brain injuries: HIC (Head Injury Criterion), peak deceleration, Face-bone damage and energy absorption. The impact energy is simulated by a drop test using a 48 mm-radius-steel hemispherical impactor, with a weight of 4.63 kg similar to that of headform J specified in AS/NZS standard. The drop-height is 1500 mm, and the linear deceleration force of the impactor is recorded and used to establish the performance of the materials. The HIC is used to predict the risk of brain injury, whereas the developed face surrogate is used to assess facial bone injury. A 5.4 m/s facial impact to the unprotected-face of a cyclist can result in the risk of severe facial bone fracture and mild brain injury. The impact test results for rigid foam protection of 40 mm thickness shows no densification (bottom out) and absorbs the impact energy without damage to the Foam-bone of the face surrogate. At 20 mm thickness, rigid polyurethane foams performed best with Foam-bone damage ranging from 15.1% to 20.5%. Other materials with thicknesses of 20 to 28 mm showed Foam-bone damage between 21.8% and 35.1%. The HIC values ranged from 267 to 522, with memory foams and expanded polystyrene foam having the lowest values. Peak deceleration ranged from 71 g to 105 g for the materials tested. It is concluded that the impact energy can be dissipated by the protection material thereby reducing the risk of severe facial injury to the protected area.展开更多
Objective:To study the relationship between the recovery of knee joint injury and traditional Chinese medicine constitution of young male amateur cyclist in Hainan.Methods:720 young male amateur cyclists were selected...Objective:To study the relationship between the recovery of knee joint injury and traditional Chinese medicine constitution of young male amateur cyclist in Hainan.Methods:720 young male amateur cyclists were selected from the tour nanli lake national wetland park cycling race and Hainan international tourism island cycling league(ding'an station)in 2019.According to the classification criteria of TCM constitutions in classification and determination of TCM constitutions revised by the Chinese association of traditional Chinese medicine in April 2009,the TCM constitutions were divided into mild constitution group and Yin deficiency constitution group.Standard Kujala knee pain questionnaire was used to evaluate knee joint function in pinyin and Yin deficiency groups.To observe the relationship between TCM constitution and functional injury and recovery degree of knee joint.Results:Body mass index(BMI)of Yin deficiency quality was lower than that of pinyin deficiency quality(P<0.01).Score of knee flexion and extension degree was higher in Yin deficiency quality 30 minutes before the race(P<0.05).The score of Yin deficiency quality was lower after 30 minutes of squatting(P<0.01)and jumping(P<0.05).The total score of knee joint was higher 5 days after the race(P<0.01).The knee flexion and extension(P<0.05)and jump(P<0.05)were higher in the pre-race and post-race scores.Conclusion:There is a high correlation between the constitution of traditional Chinese medicine and the degree of functional injury and recovery of knee joint.The degree of post-game sports injury is higher and the recovery of post-game sports injury is faster.Therefore,different sports protection schemes can be formulated for different constitutions in bicycle RACES to reduce the occurrence of sports injuries and accelerate the recovery time of sports injuries.展开更多
There is a growing global interest in developing and accommodating sustainable modes of travel.Cycling is an important mode of travel which offers powerful solutions to chronic traffic problems of congestion and emiss...There is a growing global interest in developing and accommodating sustainable modes of travel.Cycling is an important mode of travel which offers powerful solutions to chronic traffic problems of congestion and emissions.The accuracy of cyclist position is considered as an important parameter for the calculation of cyclist speed.Cyclist speed data is required for precise design of traffic control measures,for safety studies,and for sight distance analysis.Different speed measurement techniques and results are found in the literature.This study critically examines peer-reviewed studies which discuss different techniques for measuring cyclist speed.This review covers the accuracy of the measurement techniques and highlights limitations of the reviewed studies.These identified limitations are classified as:limited range of movement directions,selection of observed cyclists,seasonal variation in measurements,completeness of results reporting,reporting of equipment limitations,and measurement validation.The study summarizes previous findings of cyclist speed statistics.This study also reviews the level of automation in speed measurement.Any technique involving a human input during the field data collection or analysis to estimate cyclist speed is classified as a manual.Further categories for semiautomated and automated measurement techniques are established.Meta-analysis was conducted to test whether reported cyclist speed depends on where they were recorded;intersections or road segments.The result of the unpaired t-test showed that there is no significant difference between the means of the cyclists speed at road sections and signalized intersections at the 95%of confidence level.The study emphasizes the growing importance of the use of automated computer vision techniques for speed measurement.The paper contrasts the advantages of computer vision techniques with other measurement techniques.展开更多
基金supported by the Cultivation Program for Major Scientific Research Projects of Harbin Institute of Technology(ZDXMPY20180109).
文摘Realistic urban scene generation has been extensively studied for the sake of the development of autonomous vehicles. However, the research has primarily focused on the synthesis of vehicles and pedestrians, while the generation of cyclists is rarely presented due to its complexity. This paper proposes a perspective-aware and realistic cyclist generation method via object retrieval. Images, semantic maps, and depth labels of objects are first collected from existing datasets, categorized by class and perspective, and calculated by an algorithm newly designed according to imaging principles. During scene generation, objects with the desired class and perspective are retrieved from the collection and inserted into the background, which is then sent to the modified 2D synthesis model to generate images. This pipeline introduces a perspective computing method, utilizes object retrieval to control the perspective accurately, and modifies a diffusion model to achieve high fidelity. Experiments show that our proposal gets a 2.36 Fréchet Inception Distance, which is lower than the competitive methods, indicating a superior realistic expression ability. When these images are used for augmentation in the semantic segmentation task, the performance of ResNet-50 on the target class can be improved by 4.47%. These results demonstrate that the proposed method can be used to generate cyclists in corner cases to augment model training data, further enhancing the perception capability of autonomous vehicles and improving the safety performance of autonomous driving technology.
基金The National Key Research and Development Program of China(No.2017YFC0803902).
文摘In order to analyze the risky factors that affect vehicle-cyclist crash injury severity at the intersection area,especially the factors relating to the road users behaviors,an empirical study was conducted by collecting accident records from 2011 to 2015 from the General Estimates System.After preliminary screening,the variables were classified into 5 main categories including cyclists characteristic and behavior,drivers characteristic and behavior,vehicle characteristic,intersection condition,and time.The random parameter ordinal probit(RPOP)was used to study the significant influencing factors and corresponding heterogeneity.The results show that failing to obey traffic signals,failing to yield to right-of-way,dash and drinking before cycling can increase the injury severity for cyclists,and the corresponding fatal injury likelihoods increase by 53.2%,40.0%,86.3%,and 211.5%,respectively.Moreover,drivers inattention,speeding,going straight and left turning increase the risk of crashing for cyclists.The corresponding fatal injury likelihoods increase by 134.5%,186.5%,69.3%,and 22.7%,respectively.Other indicators such as age,gender,vehicle type,traffic signal and intersection type can also affect injury severity.
基金Projects(51208256,51178157)supported by the National Natural Science Foundation of ChinaProject(2012-K5-13)supported by the Ministry of Housing and Urban-Rural Development of China
文摘This work aimed to explore the switch tendency of bicycle use between cyclists and non-cyclists.The attitude based market segmentation approach was proposed to achieve the research objective.The filed investigations were conducted in Nanjing,China,to obtain travelers' actions and attitudes towards bicycle uses.The structural equation modeling(SEM) was used to identify the attitudinal factors indicating variables and to explore the interrelationships among them.The SEMs were developed separately for the cyclist group and the non-cyclist group.All respondents were clustered into eight distinct segments by six selected attitudinal factors.The mode switch tendency and attitude in each segment is different from others,indicating that different segments have particular policies or strategies to encourage cycling.Policy implications that best serve the potential bicycle users were discussed to reduce the number of cyclists who have high tendency to use other modes,and increase the possibility of using bicycle in the non-cyclists group with the moderate and high switch tendency.
文摘The concept of facial impact protection mask for cyclists is proposed in response to increased participation in cycling and the need for injury prevention. The research aims to develop an approach for design of facial impact protection gear to reduce the risk of severe injury. Impact test equipment and procedure, face surrogate and protection material performance criteria are developed. Three groups of protective materials – rigid crushable, semi rigid, and soft cushion foams are tested and assessed according to criteria. The criteria are linked to measures of the risk of facial and brain injuries: HIC (Head Injury Criterion), peak deceleration, Face-bone damage and energy absorption. The impact energy is simulated by a drop test using a 48 mm-radius-steel hemispherical impactor, with a weight of 4.63 kg similar to that of headform J specified in AS/NZS standard. The drop-height is 1500 mm, and the linear deceleration force of the impactor is recorded and used to establish the performance of the materials. The HIC is used to predict the risk of brain injury, whereas the developed face surrogate is used to assess facial bone injury. A 5.4 m/s facial impact to the unprotected-face of a cyclist can result in the risk of severe facial bone fracture and mild brain injury. The impact test results for rigid foam protection of 40 mm thickness shows no densification (bottom out) and absorbs the impact energy without damage to the Foam-bone of the face surrogate. At 20 mm thickness, rigid polyurethane foams performed best with Foam-bone damage ranging from 15.1% to 20.5%. Other materials with thicknesses of 20 to 28 mm showed Foam-bone damage between 21.8% and 35.1%. The HIC values ranged from 267 to 522, with memory foams and expanded polystyrene foam having the lowest values. Peak deceleration ranged from 71 g to 105 g for the materials tested. It is concluded that the impact energy can be dissipated by the protection material thereby reducing the risk of severe facial injury to the protected area.
文摘Objective:To study the relationship between the recovery of knee joint injury and traditional Chinese medicine constitution of young male amateur cyclist in Hainan.Methods:720 young male amateur cyclists were selected from the tour nanli lake national wetland park cycling race and Hainan international tourism island cycling league(ding'an station)in 2019.According to the classification criteria of TCM constitutions in classification and determination of TCM constitutions revised by the Chinese association of traditional Chinese medicine in April 2009,the TCM constitutions were divided into mild constitution group and Yin deficiency constitution group.Standard Kujala knee pain questionnaire was used to evaluate knee joint function in pinyin and Yin deficiency groups.To observe the relationship between TCM constitution and functional injury and recovery degree of knee joint.Results:Body mass index(BMI)of Yin deficiency quality was lower than that of pinyin deficiency quality(P<0.01).Score of knee flexion and extension degree was higher in Yin deficiency quality 30 minutes before the race(P<0.05).The score of Yin deficiency quality was lower after 30 minutes of squatting(P<0.01)and jumping(P<0.05).The total score of knee joint was higher 5 days after the race(P<0.01).The knee flexion and extension(P<0.05)and jump(P<0.05)were higher in the pre-race and post-race scores.Conclusion:There is a high correlation between the constitution of traditional Chinese medicine and the degree of functional injury and recovery of knee joint.The degree of post-game sports injury is higher and the recovery of post-game sports injury is faster.Therefore,different sports protection schemes can be formulated for different constitutions in bicycle RACES to reduce the occurrence of sports injuries and accelerate the recovery time of sports injuries.
文摘There is a growing global interest in developing and accommodating sustainable modes of travel.Cycling is an important mode of travel which offers powerful solutions to chronic traffic problems of congestion and emissions.The accuracy of cyclist position is considered as an important parameter for the calculation of cyclist speed.Cyclist speed data is required for precise design of traffic control measures,for safety studies,and for sight distance analysis.Different speed measurement techniques and results are found in the literature.This study critically examines peer-reviewed studies which discuss different techniques for measuring cyclist speed.This review covers the accuracy of the measurement techniques and highlights limitations of the reviewed studies.These identified limitations are classified as:limited range of movement directions,selection of observed cyclists,seasonal variation in measurements,completeness of results reporting,reporting of equipment limitations,and measurement validation.The study summarizes previous findings of cyclist speed statistics.This study also reviews the level of automation in speed measurement.Any technique involving a human input during the field data collection or analysis to estimate cyclist speed is classified as a manual.Further categories for semiautomated and automated measurement techniques are established.Meta-analysis was conducted to test whether reported cyclist speed depends on where they were recorded;intersections or road segments.The result of the unpaired t-test showed that there is no significant difference between the means of the cyclists speed at road sections and signalized intersections at the 95%of confidence level.The study emphasizes the growing importance of the use of automated computer vision techniques for speed measurement.The paper contrasts the advantages of computer vision techniques with other measurement techniques.
