Enhancing multi-scenario applicability of freeway variable speed limit control strategies using continual learning.

Variable speed limit (VSL) control benefits freeway operations through dynamic speed limit adjustment strategies for specific operation scenarios, such as traffic jams, secondary crash prevention, etc. To develop optimal strategies, deep reinforcement learning (DRL) has been employed to map the traffic operation status to speed limits with the corresponding control effects. Then, VSL control strategies were obtained based upon memories of these complex mapping relationships. However, under multi-scenario conditions, DRL trained VSL faces the challenge of performance decay, where the control strategy effects drop sharply for early trained "old scenarios". This so-called scenario forgetting problem is attributed to the fact that DRL would forget the learned old scenario mapping memories after new scenario trainings. To tackle this issue, a continual learning approach has been introduced in this study to enhance the multi-scenario applicability of VSL control strategies. Specifically, a gradient projection memory (GPM) based neural network parameter updating method was proposed to keep the mapping memories of old scenarios during new scenario trainings by imposing constraints on the direction of gradient updates for new tasks. The proposed method was evaluated using three typical freeway operation scenarios developed in the simulation platform SUMO. Experimental results showed that the continual learning approach has substantially reduced the performance decay in old scenarios by 17.76% (valued using backward transfer metrics). Furthermore, the multi-scenario VSL control strategies successfully reduced the speed standard deviation and average travel time by 28.77% and 7.25% respectively. Moreover, the generalization of the proposed continual learning based VSL approach were evaluated and discussed.

Research on Calculation Method of Radiation Response Eigenvalue of a Single-Chip Active Pixel Sensor.

In this paper, we present a calculation method for the radiation response eigenvalue based on a monolithic active pixel sensor. By comparing the statistical eigenvalues of different regions of a pixel array in bright and dark environments, the linear relationship between the statistical eigenvalues obtained by different algorithms and the radiation dose rate was studied. Additionally, a dose rate characterization method based on the analysis of the eigenvalues of the MAPS response signal was proposed. The experimental results show that in the dark background environment, the eigenvalues had a good linear response in the region of any gray value in the range of 10-30. In the color images, due to the difference in the background gray values in adjacent color regions, the radiation response signal in dark regions was confused with the image information in bright regions, resulting in the loss of response signal and affecting the analysis results of the radiation response signal. For the low dose rate radiation field, as the radiation response signal was too weak and there was background dark noise, it was necessary to accumulate frame images to obtain a sufficient response signal. For the intense radiation field, the number of response events in a single image was very high, and only two consecutive frames of image data needed to be accumulated to meet the statistical requirements. The binarization method had a good characterization effect for the radiation at a low dose rate, and the binarization processing and the total gray value statistics of the response data at a high dose rate could better characterize the radiation dose rate. The calibration experiment results show that the binarization processing method can meet the requirements of using a MAPS for wide-range detection.

Real-Time Monitoring Method for Radioactive Substances Using Monolithic Active Pixel Sensors (MAPS).

This study presents a real-time monitoring technique for radioactive substances that meets safety management needs. We studied the accumulation characteristics of radiation response signals of monolithic active pixel sensors (MAPSs) based on their response and discrimination ability to gamma (γ) photon or neutron radiation. The radiation status of the radioactive substances was determined by monitoring the accumulation data of radiation responses. As per the results, Am-Be and 252Cf radiation response signals are primarily concentrated in the range of 0-70 pixels. Response signals of 60Co and 137Cs γ-ray were concentrated in two regions; there was a peak in the region with a pixel value of less than 50, and a plateau in the region with a pixel value of more than 75. Therefore, the results are able to discriminate between spectra. Furthermore, we designed a radioactivity monitoring system that is able to examine multiple radioactive materials. Its working principle is that a change in the accumulation of radioactivity monitoring data indicates a radiation change during the last accumulation cycle. This study provides vital technical support for the long-term supervision of radioactive substances.

Strong Radiation Field Online Detection and Monitoring System with Camera.

Herein, we report the γ-ray ionizing radiation response of a commercial monolithic active-pixel sensor (MAPS) camera under strong-dose-rate irradiation with an online detection and monitoring system for strong radiation conditions. We present the first results of the distribution of three types of MAPS camera and establish a linear relationship between the average response signal and radiation dose rate in the strong-dose-rate range. There is an obvious response signal in the video frames when the camera module parameters are set to automatic, but the linear response is very poor. However, the fixed image parameters are not good at adapting to the changes of the environment and affect the quality of the video frames. A dual module online radiation detection and monitoring probe was made to carry out effective video monitoring and radiation detection at the same time. The measurement results show that the dose rate detection error is less than 5% with a dose rate in the range of 60 to 425 Gy/h, and the visible light image does not have obvious distortion, deformation, or color shift due to the interference of the radiation response event and radiation damage. Hence, the system test results show that it can be used for online detection and monitoring in a strong radiation environment.

Low dose rate γ-ray detection using a MAPS camera under a neutron radiation environment.

We present γ-ray radiation detection in a neutron radiation environment using a monolithic active pixel sensor (MAPS) camera without conversion or shielding layers. The measured output signal is the sum of the pedestal value, noise, and real radiation response signal. The sensor response shows that the MAPS camera is sensitive to neutrons and can capture a single photon. The number of pixels with a signal exceeding 100 exhibits a strong dependence on the dose rate and is the best indicator of this value. Therefore, a MAPS camera can be efficiently used as a radiation detection sensor in a robotic system, further limiting human errors in performing radiation detection in complex nuclear radiation environments.

Effect of Commercial Off-The-Shelf MAPS on γ-Ray Ionizing Radiation Response to Different Integration Times and Gains.

We report the γ-ray ionizing radiation response of commercial off-the-shelf (COTS) monolithic active-pixel sensors (MAPS) with different integration times and gains. The distribution of the eight-bit two-dimensional matrix of MAPS output frame images was studied for different parameter settings and dose rates. We present the first results of the effects of these parameters on the response of the sensor and establish a linear relationship between the average response signal and radiation dose rate in the high-dose rate range. The results show that the distribution curves can be separated into three ranges. The first range is from 0 to 24, which generates the first significant low signal peak. The second range is from 25 to 250, which shows a smooth gradient change with different integration times, gains, and dose rates. The third range is from 251 to 255, where a final peak appears, which has a relationship with integral time, gain, and dose rate. The mean pixel value shows a linear dependence on the radiation dose rate, albeit with different calibration constants depending on the integration time and gain. Hence, MAPS can be used as a radiation monitoring device with good precision.

Video Monitoring Application of CMOS 4T-PPD-APS Under γ-ray Radiation.

In this paper, we discuss the potential use of four transistor active pixel sensor (4T-APS) as a video monitor at a nuclear accident site with a high level of γ radiation. The resistance and radiation responses to γ radiation were investigated by radiation experiments using 137Cs and 60Co γ-ray sources. The radiation resistance of 4T-APS was studied by testing the mean and the maximum dark current of the sensors after irradiation. A random spatial distribution of radiation response events was observed upon analyzing these events on the video images in a given time during irradiation. The background dependence of the 4T-APS was also studied by comparing the grayscale incremental value of the images with different color and grayscale backgrounds: the radiation response events were obvious on the images with a background having a smaller grayscale value or a deeper color. Finally, the color saturation and resolution of the images were tested using a vector oscilloscope and a test card. When the total ionizing dose was less than or equal to the damage threshold, no significant performance deterioration of 4T-APS was observed in an environment with sufficient light.