SEIDM: A Safe and Efficient Intelligent Driver Model for Autonomous Driving Behavior

Yuyang Yao; Shaocheng Luo

eess.SYApr 11, 2026

SEIDM: A Safe and Efficient Intelligent Driver Model for Autonomous Driving Behavior

Authors: Yuyang Yao, Shaocheng Luo

Organizations: Department of Information and Communication Engineering, Tongji University, Shanghai, China · Department of Electrical and Computer Engineering, Duke University, Durham, NC, USA

Abstract

The Intelligent Driver Model (IDM) is a cornerstone of Adaptive Cruise Control (ACC), valued for its interpretable parameters and effectiveness in car-following behavior modeling. However, its inherent conservatism leads to prolonged stabilization and reduced traffic efficiency, which have received limited attention. In this paper, we propose SEIDM (Safe and Efficient Intelligent Driver Model), an enhanced IDM extension designed to improve traffic flow efficiency without sacrificing safety. SEIDM introduces an adaptive safety factor to dynamically modulate the impact of the safe deceleration term in acceleration decisions. This allows vehicles to follow more assertively under safe conditions while behaving more cautiously in potential hazards. Extensive urban traffic simulations show that SEIDM achieves significantly shorter stabilization spacing and faster convergence to traffic flow equilibrium, outperforming the original IDM and its variants in traffic stability and efficiency.

Figures & tables

Fig. 1: Representation of IDM. Vehicle B continuously adjusts its acceleration based on its and Vehicle A’s motion.

Fig. 2: The illustration of follower acceleration versus inter-vehicle distance across a range of relative speeds, with corresponding TTC values shown on the right axis. In this analysis, the follower’s speed is constant at 90 $\mathrm{km/h}$ , while the leader’s speed varies from $85\%$ to $100\%$ of this value. The IDM parameters from Section V-B are used.

Fig. 3: Comparison between the stabilized following distance and the desired distance in IDM. The horizontal axis represents the ratio of the stabilized speed to the desired speed $v_{0}$ . As the stabilized speed approaches $v_{0}$ , the gap between the actual and desired distances increases markedly, highlighting IDM’s growing deviation from optimal spacing at higher speeds.

Fig. 4: Illustration of SEIDM behavior. Compared to IDM (solid lines), SEIDM (dashed lines) provides stronger deceleration for quicker hazard response (Vehicle B → Vehicle A) and more aggressive acceleration to reach the desired speed efficiently under safe conditions (Vehicle C → Vehicle D).

Fig. 5: Leader speed profile.

Parameter	Variable	Value	Unit
Maximum acceleration	$a_{0}$	1.46	$\mathrm{m/s^{2}}$
Comfortable deceleration	$b_{0}$	2.0	$\mathrm{m/s^{2}}$
Desired speed	$v_{0}$	100	$\mathrm{km/h}$
Limited maximum speed	$v_{\rm max}$	95	$\mathrm{km/h}$
Safe time headway	$T$	1.6	$\mathrm{s}$
Static required distance	$s_{0}$	2	$\mathrm{m}$

TABLE I: IDM’s and SEIDM’s Parameter Configuration

$r$	$0$	$0.05$	$0.1$	$0.2$
Stabilization Spacing ( $\mathrm{m}$ )	$102.67$	$100.47$	$98.41$	$94.70$
Traffic Throughput ( $\mathrm{veh/h}$ )	$925.3$	$945.5$	$965.3$	$1003.1$
Braking Duration ( $\mathrm{s}$ )	$31.6$	$29.3$	$26.95$	$25.8$
Peak Deceleration ( $\mathrm{m/s^{2}}$ )	$-4.05$	$-4.14$	$-4.2$	$-4.31$
$r$	$0.3$	$0.4$	$0.5$	$0.6$
Stabilization Spacing ( $\mathrm{m}$ )	$91.43$	$88.53$	$85.95$	$83.64$

TABLE II: Results on the Selection of

r

Model	Spacing ( $\mathrm{m}$ )	Period ( $\mathrm{s}$ )	Traffic Throughput ( $\mathrm{veh/h}$ )
Krauss-FM [ 21 ]	$44.28$	$3.4$	$2145.44$
IDM [ 2 ]	$102.67$	$2472.4$	$925.29$
Deng et al. [ 22 ]	$102.67$	$2483.2$	$925.29$
Derbel et al. [ 1 ]	-	-	-
Dardour et al. [ 3 ]	$196.06$	$4490.6$	$484.55$
SEIDM	$83.64$	$1479.8$	$1135.8$

TABLE III: Experimental Results in Traffic Flow Stabilization Scenario

Fig. 6: Relationship among the ego velocity, the leader velocity, and the safe velocity of a vehicle in traffic flow.

Fig. 7: Relationship between average distance and simulation time across various car-following models.

Model	Final Spacing ( $\mathrm{m}$ )	Spacing Reduction ( $\mathrm{m}$ )
IDM [ 2 ]	$17.96$	$84.71$
SEIDM	$17.45$	$66.19$

TABLE IV: Experimental Results in the Consecutive Vehicle Emergency Braking Scenario

Model	Metric	0	1	2	3	4
IDM [ 2 ]	Final Spacing ( $\mathrm{m}$ )	17.98	17.98	17.98	17.97	17.97
IDM [ 2 ]	Spacing Reduction ( $\mathrm{m}$ )	84.69	84.69	84.69	84.70	84.70
SEIDM	Final Spacing ( $\mathrm{m}$ )	17.47	17.47	17.47	17.46	17.46
SEIDM	Spacing Reduction ( $\mathrm{m}$ )	66.17	66.17	66.17	66.18	66.18
Model	Metric	5	6	7	8	9
IDM [ 2 ]	Final Spacing ( $\mathrm{m}$ )	17.70	17.970	17.97	17.96	-

TABLE V: Experimental Results in the Traffic Flow Emergency Braking Scenario

Model	Stabilization Period ( $\mathrm{s}$ )	Response Time ( $\mathrm{s}$ )
IDM [ 2 ]	$2024.2$	$555.9$
SEIDM	$1276.72$	$317.516$

TABLE VI: Experimental Results in the Vehicle Insertion Scenario

Autonomous Driving

Autonomous driving develops vehicles that perceive road scenes, predict interactions, and plan and control safe movement with limited or no human input.

See all 536 papers

Learn2Drive: A neural network-based framework for socially compliant automated vehicle control
Sep 30, 2025Yuhui Liu, Samannita Halder, Shian Wang +1Intelligent Transportation Systems Autonomous Driving
Department of Civil Engineering Saint Louis University · Department of Electrical and Computer Engineering Saint Louis University · Department of Civil, Environmental, and Architectural Engineering The University of Kansas
Projected Attainable Speed Space: A Driving Efficiency Metric Connecting Instantaneous Evaluation to Travel Time
Apr 27, 2026Xiaohua Zhao, Zhaowei Huang, Chen Chen +1Autonomous Driving
Beijing Key Laboratory of Traffic Engineering, Beijing University of Technology, Beijing, 100124, P.R.China
REARL: A Closed-loop Autonomous Driving Simulation Enhancement Framework with Real Traffic Data and Large Language Models
Sep 17, 2026Xiaojun Bi, Jun Jiang, Yiwen Sun +6Language Model-Based Control Human Behavior Simulation
Minzu University of China, Beijing, China · Key Laboratory of Ethnic Language Intelligent Analysis and Security Governance of MOE, Minzu University of China, Beijing, 100081, China · Peking University, Beijing, China +4

Sep 30, 2025J/K move · Enter open · S save

Learn2Drive: A neural network-based framework for socially compliant automated vehicle control

Yuhui Liu, Samannita Halder, Shian Wang, Tianyi Li

Department of Civil Engineering Saint Louis University · Department of Electrical and Computer Engineering Saint Louis University · Department of Civil, Environmental, and Architectural Engineering The University of Kansas

Intelligent Transportation Systems Autonomous Driving Neural Network Control

Open paper AskPDF arXiv

This study introduces a novel control framework for adaptive cruise control (ACC) in automated driving, leveraging neural networks and physics-informed constraints. As automated vehicles (AVs) adopt advanced features like ACC, transportation systems are becoming increasingly intelligent and efficient. However, existing AV control strategies primarily focus on optimizing the performance of individual vehicles or platoons, often neglecting their interactions with human-driven vehicles (HVs) and the broader impact on traffic flow. This oversight can exacerbate congestion and reduce overall system efficiency. To address this critical research gap, we propose a neural network-based, socially compliant AV control framework that incorporates social value orientation (SVO). This framework enables AVs to account for their influence on HVs and traffic dynamics. By leveraging AVs as mobile traffic regulators, the proposed approach promotes adaptive driving behaviors that reduce congestion, improve traffic efficiency, and lower energy consumption. Numerical results demonstrate the effectiveness of the proposed method in adapting to varying traffic conditions, thereby enhancing system-wide efficiency. Specifically, when the AV's control mode shifts from prioritizing its own energy conservation to optimizing collective traffic flow efficiency, the controlled AV proactively adapts its acceleration profile. This prosocial behavior yields at least a 38.39% improvement in the average speed of downstream vehicles and effectively dampens traffic oscillations, demonstrating significant enhancements in system-wide dynamics. The implementation code is available https://github.com/lilab2024/Learn2Drive-SVO_v1.git.

Abstract

Autonomous Driving

Explore similar work

Learn2Drive: A neural network-based framework for socially compliant automated vehicle control

Projected Attainable Speed Space: A Driving Efficiency Metric Connecting Instantaneous Evaluation to Travel Time

REARL: A Closed-loop Autonomous Driving Simulation Enhancement Framework with Real Traffic Data and Large Language Models