This article studies the effective traffic signal control problem of multiple intersections in a city-level traffic system.A novel regional multi-agent cooperative reinforcement learning algorithm called RegionSTLight...This article studies the effective traffic signal control problem of multiple intersections in a city-level traffic system.A novel regional multi-agent cooperative reinforcement learning algorithm called RegionSTLight is proposed to improve the traffic efficiency.Firstly a regional multi-agent Q-learning framework is proposed,which can equivalently decompose the global Q value of the traffic system into the local values of several regions Based on the framework and the idea of human-machine cooperation,a dynamic zoning method is designed to divide the traffic network into several strong-coupled regions according to realtime traffic flow densities.In order to achieve better cooperation inside each region,a lightweight spatio-temporal fusion feature extraction network is designed.The experiments in synthetic real-world and city-level scenarios show that the proposed RegionS TLight converges more quickly,is more stable,and obtains better asymptotic performance compared to state-of-theart models.展开更多
Recently, various control methods represented by proportional-integral-derivative （PID） control are used for robotic control. To cope with the requirements for high response and precision, advanced feedforward contr...Recently, various control methods represented by proportional-integral-derivative （PID） control are used for robotic control. To cope with the requirements for high response and precision, advanced feedforward controllers such as gravity compensator, Coriolis/centrifugal force compensator and friction compensators have been built in the controller. Generally, it causes heavy computational load when calculating the compensating value within a short sampling period. In this paper, integrated recurrent neural networks are applied as a feedforward controller for PUMA560 manipulator. The feedforward controller works instead of gravity and Coriolis/centrifugal force compensators. In the learning process of the neural network by using back propagation algorithm, the learning coefficient and gain of sigmoid function are tuned intuitively and empirically according to teaching signals. The tuning is complicated because it is being conducted by trial and error. Especially, when the scale of teaching signal is large, the problem becomes crucial. To cope with the problem which concerns the learning performance, a simple and adaptive learning technique for large scale teaching signals is proposed. The learning techniques and control effectiveness are evaluated through simulations using the dynamic model of PUMA560 manipulator.展开更多
A control system for correction mechanisms through the whole trajectory is proposed based on the principle of one-dimensional trajectory correction projectile. Digital signal processing( DSP) is utilized as the core c...A control system for correction mechanisms through the whole trajectory is proposed based on the principle of one-dimensional trajectory correction projectile. Digital signal processing( DSP) is utilized as the core controller and gobal positioning system( GPS) is used to measure trajectory parameters to meet the requirements of calculating ballistics and system functions. Firstly,the hardware,mainly including communication module,ballistic calculation module,boosting& detonating module and data storage module,is designed. Secondly,the supporting software is developed based on the communication protocols of GPS and the workflow of control system. Finally,the feasibility and the reliability of the control system are verified through dynamic tests in a car and live firing experiments. The system lays a foundation for the research on trajectory correction projectile for the whole trajectory.展开更多
Updates to traffic signal timing plans are expected to either improve operations or mitigate the effects of increased volumes. Longitudinal before-after studies are important when validating changes to traffic signal ...Updates to traffic signal timing plans are expected to either improve operations or mitigate the effects of increased volumes. Longitudinal before-after studies are important when validating changes to traffic signal systems, but they have historically required field data collection as well as deployment of extensive detection and communication equipment. These infrastructure-based techniques are costly and hard to scale. This study utilizes commercially available connected vehicle (CV) trajectory data to assess the change in performance between August 2020 and August 2021 on a 22-intersection corridor associated with the implementation of a semi-automated adaptive control system. Approximately 1 million trajectories and 13.5 million GPS points are analyzed for weekdays in August 2020 and August 2021. The vehicle trajectory data is used to compute corridor travel times and linear referenced relative to the far side of each intersection to generate Purdue Probe Diagrams (PPD). Using the PPDs, operational measurements such as arrivals on green (AOG), split failures (SF), and downstream blockage (DSB) are calculated. Additionally, traditional Highway Capacity Manual (HCM) level of service (LOS) is estimated. Even though there was a 35% increase in annual average daily traffic (AADT), the weighted average vehicle delay only increased by two seconds, LOS did not change, AOG improved by 1%, and SF and DSB remained the same. Based on the small changes in operational performance and considering the increase in traffic volume it is concluded that the implementation of the semi-automated adaptive control system had a significant positive impact in the corridor. The presented framework can be utilized by agencies to use CV data to perform before-after studies to evaluate the impact of signal timing plan changes. The presented methodology can be applied to any location where CV trajectory data is available.展开更多
基金supported by the National Science and Technology Major Project (2021ZD0112702)the National Natural Science Foundation (NNSF)of China (62373100,62233003)the Natural Science Foundation of Jiangsu Province of China (BK20202006)。
文摘This article studies the effective traffic signal control problem of multiple intersections in a city-level traffic system.A novel regional multi-agent cooperative reinforcement learning algorithm called RegionSTLight is proposed to improve the traffic efficiency.Firstly a regional multi-agent Q-learning framework is proposed,which can equivalently decompose the global Q value of the traffic system into the local values of several regions Based on the framework and the idea of human-machine cooperation,a dynamic zoning method is designed to divide the traffic network into several strong-coupled regions according to realtime traffic flow densities.In order to achieve better cooperation inside each region,a lightweight spatio-temporal fusion feature extraction network is designed.The experiments in synthetic real-world and city-level scenarios show that the proposed RegionS TLight converges more quickly,is more stable,and obtains better asymptotic performance compared to state-of-theart models.
基金supported by Grant-in-Aid for Scientific Research(C) (No. 20560248) of Japan
文摘Recently, various control methods represented by proportional-integral-derivative （PID） control are used for robotic control. To cope with the requirements for high response and precision, advanced feedforward controllers such as gravity compensator, Coriolis/centrifugal force compensator and friction compensators have been built in the controller. Generally, it causes heavy computational load when calculating the compensating value within a short sampling period. In this paper, integrated recurrent neural networks are applied as a feedforward controller for PUMA560 manipulator. The feedforward controller works instead of gravity and Coriolis/centrifugal force compensators. In the learning process of the neural network by using back propagation algorithm, the learning coefficient and gain of sigmoid function are tuned intuitively and empirically according to teaching signals. The tuning is complicated because it is being conducted by trial and error. Especially, when the scale of teaching signal is large, the problem becomes crucial. To cope with the problem which concerns the learning performance, a simple and adaptive learning technique for large scale teaching signals is proposed. The learning techniques and control effectiveness are evaluated through simulations using the dynamic model of PUMA560 manipulator.
文摘A control system for correction mechanisms through the whole trajectory is proposed based on the principle of one-dimensional trajectory correction projectile. Digital signal processing( DSP) is utilized as the core controller and gobal positioning system( GPS) is used to measure trajectory parameters to meet the requirements of calculating ballistics and system functions. Firstly,the hardware,mainly including communication module,ballistic calculation module,boosting& detonating module and data storage module,is designed. Secondly,the supporting software is developed based on the communication protocols of GPS and the workflow of control system. Finally,the feasibility and the reliability of the control system are verified through dynamic tests in a car and live firing experiments. The system lays a foundation for the research on trajectory correction projectile for the whole trajectory.
文摘Updates to traffic signal timing plans are expected to either improve operations or mitigate the effects of increased volumes. Longitudinal before-after studies are important when validating changes to traffic signal systems, but they have historically required field data collection as well as deployment of extensive detection and communication equipment. These infrastructure-based techniques are costly and hard to scale. This study utilizes commercially available connected vehicle (CV) trajectory data to assess the change in performance between August 2020 and August 2021 on a 22-intersection corridor associated with the implementation of a semi-automated adaptive control system. Approximately 1 million trajectories and 13.5 million GPS points are analyzed for weekdays in August 2020 and August 2021. The vehicle trajectory data is used to compute corridor travel times and linear referenced relative to the far side of each intersection to generate Purdue Probe Diagrams (PPD). Using the PPDs, operational measurements such as arrivals on green (AOG), split failures (SF), and downstream blockage (DSB) are calculated. Additionally, traditional Highway Capacity Manual (HCM) level of service (LOS) is estimated. Even though there was a 35% increase in annual average daily traffic (AADT), the weighted average vehicle delay only increased by two seconds, LOS did not change, AOG improved by 1%, and SF and DSB remained the same. Based on the small changes in operational performance and considering the increase in traffic volume it is concluded that the implementation of the semi-automated adaptive control system had a significant positive impact in the corridor. The presented framework can be utilized by agencies to use CV data to perform before-after studies to evaluate the impact of signal timing plan changes. The presented methodology can be applied to any location where CV trajectory data is available.
