Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334622
Anwesha Das, Joris IJsselmuiden, Gijs Dubbelman
Autonomous vehicles are dependent on High Definition (HD) maps. The process of generating and updating these maps is slow, expensive, and not scalable for the whole world. Crowdsourcing vehicle sensor data to generate and update maps is a solution to the problem. In this paper, we propose and evaluate an end-to-end pose-graph optimization-based mapping framework using crowdsourced vehicle data. The in-vehicle data acquisition framework and the cloud-based mapping framework that fuses data from a consumer-grade Global Navigation Satellite System (GNSS) receiver, an odometry sensor, and a stereo camera is described in detail. We focus on using stereo image pairs for loop-closure detection to combine crowdsourced data from different sessions that are affected by GNSS biases. We evaluate our framework on a data-set of more than 180 km recorded around the Eindhoven area. After the map generation process, the results exhibit a 56.23% improvement in maximum offset error and a 24.39% improvement in precision around the loop-closure area.
{"title":"Pose-graph based Crowdsourced Mapping Framework","authors":"Anwesha Das, Joris IJsselmuiden, Gijs Dubbelman","doi":"10.1109/CAVS51000.2020.9334622","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334622","url":null,"abstract":"Autonomous vehicles are dependent on High Definition (HD) maps. The process of generating and updating these maps is slow, expensive, and not scalable for the whole world. Crowdsourcing vehicle sensor data to generate and update maps is a solution to the problem. In this paper, we propose and evaluate an end-to-end pose-graph optimization-based mapping framework using crowdsourced vehicle data. The in-vehicle data acquisition framework and the cloud-based mapping framework that fuses data from a consumer-grade Global Navigation Satellite System (GNSS) receiver, an odometry sensor, and a stereo camera is described in detail. We focus on using stereo image pairs for loop-closure detection to combine crowdsourced data from different sessions that are affected by GNSS biases. We evaluate our framework on a data-set of more than 180 km recorded around the Eindhoven area. After the map generation process, the results exhibit a 56.23% improvement in maximum offset error and a 24.39% improvement in precision around the loop-closure area.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"99 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126103290","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334562
Samuel Thornton, S. Dey
Emerging Vehicle-to-Everything (V2X) technologies promise to improve the perception of streets by enabling data sharing like camera views between multiple vehicles. However, to ensure accuracy of such enhanced perception, the problem of vehicle matching becomes important; the goal of a vehicle matching system is to identify if images of vehicles seen by different cameras correspond to the same vehicle. Such a system is necessary to avoid duplicate detections for a vehicle seen by multiple cameras and to avoid detections being discarded due to a false match being made. One of the most challenging scenarios in vehicle matching is when the camera positions have very large viewpoint differences, as will commonly be the case when the cameras are in geographically separate locations like in vehicles and street infrastructure. In these scenarios, traditional handcrafted features will not be sufficient to create these correspondences due to the lack of common visual features. In this paper we will examine the performance of random forests and neural networks as classifiers for both learned features and high level visual features when used for this vehicles matching problem. Additionally, a novel dataset of vehicles from cameras with very large viewpoint differences was recorded to validate our method; our preliminary results achieve high classification accuracy with low inference time which shows the feasibility of a real time vehicle matching system.
{"title":"Machine Learning Techniques for Vehicle Matching with Non-Overlapping Visual Features","authors":"Samuel Thornton, S. Dey","doi":"10.1109/CAVS51000.2020.9334562","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334562","url":null,"abstract":"Emerging Vehicle-to-Everything (V2X) technologies promise to improve the perception of streets by enabling data sharing like camera views between multiple vehicles. However, to ensure accuracy of such enhanced perception, the problem of vehicle matching becomes important; the goal of a vehicle matching system is to identify if images of vehicles seen by different cameras correspond to the same vehicle. Such a system is necessary to avoid duplicate detections for a vehicle seen by multiple cameras and to avoid detections being discarded due to a false match being made. One of the most challenging scenarios in vehicle matching is when the camera positions have very large viewpoint differences, as will commonly be the case when the cameras are in geographically separate locations like in vehicles and street infrastructure. In these scenarios, traditional handcrafted features will not be sufficient to create these correspondences due to the lack of common visual features. In this paper we will examine the performance of random forests and neural networks as classifiers for both learned features and high level visual features when used for this vehicles matching problem. Additionally, a novel dataset of vehicles from cameras with very large viewpoint differences was recorded to validate our method; our preliminary results achieve high classification accuracy with low inference time which shows the feasibility of a real time vehicle matching system.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130524965","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334596
Sina Alighanbari, N. Azad
Connectivity and automation enable vehicles to transfer crucial driving data and information to improve performance and safety. This paper proposes a nonlinear model predictive control (NMPC) control approach to address the problem of decentralized coordination of vehicles at roundabouts. A priority calculation logic is proposed and its performance is tested for different scenarios. We use simulations to test the controller and the Toyota Prius PHEV high-fidelity model is used in this paper for simulations. Simulation results show the proposed approach can determine priorities and improve performance. Also, results show that the addition of energy economy to the performance index can improve the fuel consumption of the vehicle. One of the major concerns in designing a controller for automotive applications is real-time implementation. The results of hardware-in-the-loop experiments show the real-time implementation of the controller.
{"title":"Multi-Vehicle Coordination and Real-time Control of Connected and Automated Vehicles at Roundabouts","authors":"Sina Alighanbari, N. Azad","doi":"10.1109/CAVS51000.2020.9334596","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334596","url":null,"abstract":"Connectivity and automation enable vehicles to transfer crucial driving data and information to improve performance and safety. This paper proposes a nonlinear model predictive control (NMPC) control approach to address the problem of decentralized coordination of vehicles at roundabouts. A priority calculation logic is proposed and its performance is tested for different scenarios. We use simulations to test the controller and the Toyota Prius PHEV high-fidelity model is used in this paper for simulations. Simulation results show the proposed approach can determine priorities and improve performance. Also, results show that the addition of energy economy to the performance index can improve the fuel consumption of the vehicle. One of the major concerns in designing a controller for automotive applications is real-time implementation. The results of hardware-in-the-loop experiments show the real-time implementation of the controller.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134417922","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334684
Waheeda Jabbar, R. Malaney, Shihao Yan
In vehicular networks, vehicle claimed positions should be independently verified to help protect the wider network against location-spoofing attacks. In this work, we propose a new solution to the problem of location verification using the Cramer-Rao Bound (CRB) on location accuracy. Compared to known-optimal solutions, our technique has the advantage that it does not depend on a priori information on the probability of any vehicle being malicious. To analyze the performance of our new solution, we compare its operation, under Received Signal Strength (RSS) inputs, with a known optimal solution for this problem that assumes the probability of a vehicle being malicious is known. The results show that our new solution provides close to optimal performance over a wide range of anticipated channel conditions. Our solution is simple to deploy and can easily be integrated into a host of vehicular applications that use location information as an input.
{"title":"Location Information Verification in Future Vehicular Networks","authors":"Waheeda Jabbar, R. Malaney, Shihao Yan","doi":"10.1109/CAVS51000.2020.9334684","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334684","url":null,"abstract":"In vehicular networks, vehicle claimed positions should be independently verified to help protect the wider network against location-spoofing attacks. In this work, we propose a new solution to the problem of location verification using the Cramer-Rao Bound (CRB) on location accuracy. Compared to known-optimal solutions, our technique has the advantage that it does not depend on a priori information on the probability of any vehicle being malicious. To analyze the performance of our new solution, we compare its operation, under Received Signal Strength (RSS) inputs, with a known optimal solution for this problem that assumes the probability of a vehicle being malicious is known. The results show that our new solution provides close to optimal performance over a wide range of anticipated channel conditions. Our solution is simple to deploy and can easily be integrated into a host of vehicular applications that use location information as an input.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"145 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124999867","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334578
Robert Lugner, Daniel Vriesman, Maximilian Inderst, G. Sequeira, Niyathipriya Pasupuleti, A. Zimmer, T. Brandmeier
Vehicle safety is an enabler of Automated Driving. The combination of active and passive vehicle safety can further increase the safety level of vehicle occupants. With integrated safety systems predicting inevitable crashes and the corresponding crash constellation, the activation of irreversible restraint systems like airbags will allow better crash mitigation and new interior concepts. One requirement is a comprehensive methodology to ensure the correct detection of the current traffic situation, the involved vehicles, and the collision inevitability. This paper presents a novel approach for crash evaluation in the pre-crash phase based on sensor fusion using camera and LiDAR for bullet vehicle detection in combination with physical motion-model-based collision detection. Urban intersection scenarios with typically severe side crashes are investigated using this methodology. The presented method can also be applied to investigate other traffic scenarios. One focus of this paper is the effect of sensor tolerances, which lead to inaccurate object data on the prediction of the inevitability of the crash. The analysis proves the potential of preemptive activation of airbag systems.
{"title":"Evaluation of Sensor Tolerances and Inevitability for Pre-Crash Safety Systems in Real Case Scenarios","authors":"Robert Lugner, Daniel Vriesman, Maximilian Inderst, G. Sequeira, Niyathipriya Pasupuleti, A. Zimmer, T. Brandmeier","doi":"10.1109/CAVS51000.2020.9334578","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334578","url":null,"abstract":"Vehicle safety is an enabler of Automated Driving. The combination of active and passive vehicle safety can further increase the safety level of vehicle occupants. With integrated safety systems predicting inevitable crashes and the corresponding crash constellation, the activation of irreversible restraint systems like airbags will allow better crash mitigation and new interior concepts. One requirement is a comprehensive methodology to ensure the correct detection of the current traffic situation, the involved vehicles, and the collision inevitability. This paper presents a novel approach for crash evaluation in the pre-crash phase based on sensor fusion using camera and LiDAR for bullet vehicle detection in combination with physical motion-model-based collision detection. Urban intersection scenarios with typically severe side crashes are investigated using this methodology. The presented method can also be applied to investigate other traffic scenarios. One focus of this paper is the effect of sensor tolerances, which lead to inaccurate object data on the prediction of the inevitability of the crash. The analysis proves the potential of preemptive activation of airbag systems.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126341399","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334603
Monish Gogri, Maike Hartstern, W. Stork, T. Winsel
This paper proposes a methodology for determining strategically bundled relevant test scenarios for the simulation- based evaluation of sensor constellations. This is achieved by gogri identification of important use cases for the autonomous driving operation, (b) the conversion of these use cases into Regions of Interests (ROIs) around the vehicle along with (c) a definition of a critical index (CI) for each of these regions and (d) a procedure to derive crucial scenarios and (e) categorise them into scenario families. The derived test scenarios help to optimise the field of view of the sensor constellation for the most important regions around the ego vehicle. The novelty lies in its independence from traditional methods of deriving test scenarios and its capability of providing targeted feedback to improve the sensor constellation at the identified pain points. The test scenario families can reduce the development time of highly automated vehicles by providing virtual testing of the sensor constellation performance in the vehicle concept phase.
{"title":"A Methodology to Determine Test Scenarios for Sensor Constellation Evaluations","authors":"Monish Gogri, Maike Hartstern, W. Stork, T. Winsel","doi":"10.1109/CAVS51000.2020.9334603","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334603","url":null,"abstract":"This paper proposes a methodology for determining strategically bundled relevant test scenarios for the simulation- based evaluation of sensor constellations. This is achieved by gogri identification of important use cases for the autonomous driving operation, (b) the conversion of these use cases into Regions of Interests (ROIs) around the vehicle along with (c) a definition of a critical index (CI) for each of these regions and (d) a procedure to derive crucial scenarios and (e) categorise them into scenario families. The derived test scenarios help to optimise the field of view of the sensor constellation for the most important regions around the ego vehicle. The novelty lies in its independence from traditional methods of deriving test scenarios and its capability of providing targeted feedback to improve the sensor constellation at the identified pain points. The test scenario families can reduce the development time of highly automated vehicles by providing virtual testing of the sensor constellation performance in the vehicle concept phase.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116713617","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334587
Kenan Softić, Haris Šikić, Amar Civgin, G. Stettinger, D. Watzenig
A reliable and precise model of the environment is of the highest importance for autonomous vehicles. Occupancy grids are a well-known approach for environment modeling and are a crucial part of multiple autonomous driving functionalities. The standard method is to use a single 2D occupancy grid to model the environment using nonground points. In this paper, we propose a decentralized occupancy grid filtering chain (pipeline) where a high-density 64-layer LiDAR provided the input to our pipeline. Our approach enables us to obtain detailed 2D and 3D models of the environment simultaneously. The pipeline was validated on different scenarios in both simulation and real world. The performance of the designed occupancy grid pipeline was evaluated by the proposed key performance indicators (KPIs) based on accuracy. The results have shown that the approach was able to extract free space information with a high degree of accuracy, while reducing the size of the unobserved area in the grid compared to the standard methods and achieving real-time performance.
{"title":"Validation and testing of the decentralized architecture for the occupancy grid filtering pipeline","authors":"Kenan Softić, Haris Šikić, Amar Civgin, G. Stettinger, D. Watzenig","doi":"10.1109/CAVS51000.2020.9334587","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334587","url":null,"abstract":"A reliable and precise model of the environment is of the highest importance for autonomous vehicles. Occupancy grids are a well-known approach for environment modeling and are a crucial part of multiple autonomous driving functionalities. The standard method is to use a single 2D occupancy grid to model the environment using nonground points. In this paper, we propose a decentralized occupancy grid filtering chain (pipeline) where a high-density 64-layer LiDAR provided the input to our pipeline. Our approach enables us to obtain detailed 2D and 3D models of the environment simultaneously. The pipeline was validated on different scenarios in both simulation and real world. The performance of the designed occupancy grid pipeline was evaluated by the proposed key performance indicators (KPIs) based on accuracy. The results have shown that the approach was able to extract free space information with a high degree of accuracy, while reducing the size of the unobserved area in the grid compared to the standard methods and achieving real-time performance.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126419050","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334679
Baik Hoh, Seyhan Uçar, Pratham Oza, Chinmaya Patnayak, K. Oguchi
Vehicles are getting more equipped with sensors and driver assistant systems. However, neither these technological advances nor traditional traffic enforcement systems are sufficient in protecting commuters from misbehaving drivers such as aggressive, distracted, and drunken drivers. That is why we have not observed any substantial improvement in road safety and driving experience in recent years despite those technological advances. Being motivated by the success of reputation systems (i.e., How’s My Driving (HMD), eBay, and Wikipedia), we present the concept of Collaborative On-Road Reputation (CORR) system and discuss the potential benefits and challenges ahead when we expand CORR to all vehicles. We focus on how to identify the anomalous driving behavior and propose a cooperative anomaly detection method where nearby connected vehicles collaborate to surface the anomalous driving behavior. Through extensive simulations, we demonstrate that CORR can identify the anomalous driving behavior by about 75% accuracy under a certain level of connected vehicle penetration rates.
{"title":"CORR: Collaborative On-Road Reputation","authors":"Baik Hoh, Seyhan Uçar, Pratham Oza, Chinmaya Patnayak, K. Oguchi","doi":"10.1109/CAVS51000.2020.9334679","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334679","url":null,"abstract":"Vehicles are getting more equipped with sensors and driver assistant systems. However, neither these technological advances nor traditional traffic enforcement systems are sufficient in protecting commuters from misbehaving drivers such as aggressive, distracted, and drunken drivers. That is why we have not observed any substantial improvement in road safety and driving experience in recent years despite those technological advances. Being motivated by the success of reputation systems (i.e., How’s My Driving (HMD), eBay, and Wikipedia), we present the concept of Collaborative On-Road Reputation (CORR) system and discuss the potential benefits and challenges ahead when we expand CORR to all vehicles. We focus on how to identify the anomalous driving behavior and propose a cooperative anomaly detection method where nearby connected vehicles collaborate to surface the anomalous driving behavior. Through extensive simulations, we demonstrate that CORR can identify the anomalous driving behavior by about 75% accuracy under a certain level of connected vehicle penetration rates.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122249251","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334640
M. Shaqfeh, Salah Hessien, E. Serpedin
Real-time traffic information can be utilized to enhance the efficiency of transportation networks by dynamically updating routing plans to mitigate traffic jams. However, an interesting question is whether the network coordinator should broadcast real-time traffic information to all network users or communicate it selectively to some of them. Which option enhances the network efficiency more?In this work, we demonstrate using simulation experiments that sharing real-time traffic information with all network users is sub-optimal, and it is actually better to share the information with a majority subset of the total population in order to improve the overall network performance. This result is valid under the assumption that each network user decides it’s route to destination locally.
{"title":"Utility of Traffic Information in Dynamic Routing: Is Sharing Information Always Useful?","authors":"M. Shaqfeh, Salah Hessien, E. Serpedin","doi":"10.1109/CAVS51000.2020.9334640","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334640","url":null,"abstract":"Real-time traffic information can be utilized to enhance the efficiency of transportation networks by dynamically updating routing plans to mitigate traffic jams. However, an interesting question is whether the network coordinator should broadcast real-time traffic information to all network users or communicate it selectively to some of them. Which option enhances the network efficiency more?In this work, we demonstrate using simulation experiments that sharing real-time traffic information with all network users is sub-optimal, and it is actually better to share the information with a majority subset of the total population in order to improve the overall network performance. This result is valid under the assumption that each network user decides it’s route to destination locally.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127743662","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}
Pub Date : 2020-11-01DOI: 10.1109/CAVS51000.2020.9334636
Mohsen Shirpour, S. Beauchemin, M. Bauer
The direction of a vehicle driver’s visual attention plays an essential role in the research on Advanced Driving Assistance Systems (ADAS) and autonomous vehicles. How a driver monitors the surrounding environment is at least partially descriptive of the driver’s situational awareness. While driver gaze is not explicitly related to head pose due to the interplay between head and eye movements, it may still provide an approximation of the visual attention that is sufficiently accurate for many applications. In this research, we propose a probabilistic method for describing the visual attention of drivers. This method applies a Gaussian Process Regression (GPR) technique that estimates the probability of the driver gaze direction, given head pose. We evaluate our model on real data collected during drives with an experimental vehicle in urban and suburban areas. Our experimental results show that 82.5% of drivers’ gaze lies within the 95% confidence interval predicted by our framework.
{"title":"A Probabilistic Model for Visual Driver Gaze Approximation from Head Pose Estimation","authors":"Mohsen Shirpour, S. Beauchemin, M. Bauer","doi":"10.1109/CAVS51000.2020.9334636","DOIUrl":"https://doi.org/10.1109/CAVS51000.2020.9334636","url":null,"abstract":"The direction of a vehicle driver’s visual attention plays an essential role in the research on Advanced Driving Assistance Systems (ADAS) and autonomous vehicles. How a driver monitors the surrounding environment is at least partially descriptive of the driver’s situational awareness. While driver gaze is not explicitly related to head pose due to the interplay between head and eye movements, it may still provide an approximation of the visual attention that is sufficiently accurate for many applications. In this research, we propose a probabilistic method for describing the visual attention of drivers. This method applies a Gaussian Process Regression (GPR) technique that estimates the probability of the driver gaze direction, given head pose. We evaluate our model on real data collected during drives with an experimental vehicle in urban and suburban areas. Our experimental results show that 82.5% of drivers’ gaze lies within the 95% confidence interval predicted by our framework.","PeriodicalId":409507,"journal":{"name":"2020 IEEE 3rd Connected and Automated Vehicles Symposium (CAVS)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132223298","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：