An optimal control-based vehicle speed guidance strategy to improve traffic safety and efficiency against freeway jam waves

Analysis on congestion mechanism of CAVs around traffic accident zones

Unexpected traffic accidents cause traffic congestion and aggravate the unsafe situation on the roadways. Reducing the impact of such congestion by introducing Connected and Autonomous Vehicles (CAVs) into the traditional traffic flow is possible. It requires estimating the incident's duration and analyzing the incident's impact area to determine the appropriate strategy. To guide the driver in making efficient and accurate judgments and avoiding secondary traffic congestion, the Cooperative Adaptive Cruise Control (CACC) model with dynamic safety distance and the Intelligent Driver Model (IDM) based on the safety potential field theory are introduced to build the evolution model of accidental traffic congestion under diversion interference and non-interference. The Huatao Interchange section of the Inner Ring Highway in the Banan District of Chongqing, China, was selected as the test section for simulating mixed traffic flow under different CAVs permeability (Pc). The relationship between the evacuation time, evacuation traffic volume, and the accident impact degree index (including the farthest queue length and accident duration) under the diversion intervention scenario was analyzed, respectively. The results of the study indicate that the higher the penetration of CAVs, the more significant the improvement in traffic flow occupancy, flow, and speed. Diversion interventions reduce congestion, about 50 % of the duration without interventions, when Pc ≤ 80 %. The traffic volume of diversion interference is non-linearly positively correlated with the maximum queue length, and the earlier the interference time, the stronger the positive correlation. The negative correlation between the interference time and queue length is weak at low evacuation traffic volume. With the increase in evacuation traffic volume, the influence of evacuation time on queue length becomes stronger. The maximum queue length value interval under different conditions is [348 m, 3140 m], and the shortest evacuation time is [1649 s, 2834 s]. The traffic flow data obtained from the simulation are imported into the episodic traffic congestion evolution model. The congestion evaluation indexes are calculated under non-interference and interference measures and compared with the simulation results. The maximum relative error is within 5.38 %. The results can be of great significance in relieving congestion caused by traffic accidents and promptly restoring road capacity.

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.

A cooperative control method for safer on-ramp merging process in heterogeneous traffic flow

The on-ramp area is a high-risk conflict zone where traffic accidents frequently occur. Connected and automated vehicles (CAVs) have the potential to enhance the safety of the merging process through appropriate cooperative control methods. This paper proposes a cooperative control method for safer on-ramp merging processes in heterogeneous traffic flow. Firstly, the gap selection process of ramp vehicles is described, thus all feasible virtual platoon results can be summarized. Next, the vehicle bond (VB) is used to describe the connection mode between vehicles within the virtual platoon. A two-layered gap selection function is proposed to ensure a safer merging process. The first layer aims to minimize the number of empty VBs, while the second layer considers fairness with respect to delay. To evaluate the control effectiveness, time exposed time-to-collision (TET), cumulative risk (CR), and conflict-potential mergence ratio (CPMR) are selected as the safety evaluation indicators. The simulation results show that the gap selection control moves the merging positions of ramp vehicles forward, resulting less risk of merging. It significantly enhances the safety of on-ramp merging without compromising traffic efficiency. At a flow rate of 650 veh/h for both the mainline and ramp, and a CAV penetration rate of 0.1-0.9, the gap selection control group achieves a decrease rate of about 0.3-0.6 for average TET and CR compared to the non-control group. In the pure CAV environment, the decrease rate can reach about 0.9. Sensitivity analysis indicates that the gap selection control is effective across varying flow rates and steady speeds. The optimal control effect is achieved when the length of the communication area ranges from 100 to 200 m.