Decision Making and Ride Comfort Evaluation Based on Lane-Change Prediction for Autonomous Vehicles

Mohamed Manzour; Aditya Kumar; Rubén Izquierdo; Miguel Ángel Sotelo

cs.ROOct 21, 2026

Decision Making and Ride Comfort Evaluation Based on Lane-Change Prediction for Autonomous Vehicles

Authors: Mohamed Manzour, Aditya Kumar, Rubén Izquierdo, Miguel Ángel Sotelo

Organizations: Computer Engineering Department, University of Alcalá, Madrid, Spain

Abstract

Predicting surrounding vehicles' lane-change intentions enables autonomous vehicles to anticipate potential hazards and respond proactively rather than relying solely on reactive maneuvers. This study investigates the benefits of integrating lane-change prediction, based on knowledge graph embeddings and Bayesian inference, into autonomous vehicle decision-making. Three driving strategies are evaluated in risky merging scenarios simulated in CARLA: reactive driving, prediction-based longitudinal action, and prediction-based lateral action. The proposed prediction model anticipates the surrounding vehicle's lane change approximately 3-4 seconds before its occurrence, allowing the ego vehicle to adjust its behavior accordingly. In the representative simulation records, reactive driving results in a collision accompanied by aggressive deceleration. In contrast, prediction-based longitudinal action initiates early braking to establish a safe gap, while prediction-based lateral action performs a lane change when a safe adjacent lane is available. Both predictive strategies avoid collisions in the evaluated records. Compared with reactive driving, peak deceleration before impact is reduced by 31.5% with longitudinal action and 91.0% with lateral action. Additionally, an ISO 2631-1-based horizontal vibration estimate is used to provide a ride comfort assessment. Overall, the results highlight the potential of lane-change prediction to support safer and smoother driving through proactive decision-making, with additional insights into ride comfort provided by vibration-based estimates.

Figures & tables

Figure 1: The white target vehicle, following the black preceding vehicle, is predicted to merge into the grey ego vehicle’s lane. Red: collision; blue: braking; green: lateral avoidance. Transparent vehicles indicate future positions.

Figure 2: Prediction stages: convert measurements into descriptive categories, learn graph embeddings, then use Bayesian inference to predict the maneuver. Illustrations are schematic.

Figure 3: Longitudinal velocity profiles. The reactive vehicle experiences a sudden speed reduction following impact, while anticipatory braking produces a gradual deceleration and lateral avoidance preserves forward motion. Upper: full range; lower: zoomed view.

Figure 4: Lateral velocity profiles. The avoidance maneuver (blue) introduces lateral movement toward the adjacent lane, whereas the reactive vehicle exhibits a lateral disturbance associated with the collision. Upper: full range; lower: zoomed view.

Figure 5: Longitudinal acceleration profiles. The reactive response produces a sharp deceleration at impact, while early braking distributes deceleration over a longer interval. Lateral avoidance requires substantially less longitudinal braking. Upper: full range; lower: zoomed view.

Figure 6: Lateral acceleration profiles. Lane-change execution introduces lateral acceleration as the vehicle enters the adjacent lane. Upper: full range; lower: zoomed view.

Figure 7: Longitudinal jerk profiles. The largest variation occurs during the reactive collision, whereas anticipatory actions reduce abrupt changes in longitudinal acceleration. Upper: full range; lower: zoomed view.

Figure 8: Lateral jerk profiles. The avoidance maneuver introduces acceleration variations associated with lateral movement, while the reactive collision produces a clear spike. Upper: full range; lower: zoomed view.

Metric	Reactive	Anticipatory braking	Lateral avoidance
Peak deceleration (m/s 2 )	6.427	4.406	0.578
Peak $\|a_{y}\|$ (m/s 2 )	0.131	0.130	0.530
Peak $\|j_{x}\|$ (m/s 3 )	11.853	4.116	1.087
Peak $\|j_{y}\|$ (m/s 3 )	0.275	0.279	0.882
$A_{x,\mathrm{RMS}}$ (m/s 2 )	3.366	2.658	0.230
$A_{y,\mathrm{RMS}}$ (m/s 2 )	0.037	0.037	0.388

Table 1: Comparison of unweighted longitudinal and lateral acceleration and jerk during the three seconds preceding the reactive collision. Collision-induced motion is excluded from the table but included in the figures for visualization.

Figure 9: Pre-impact RMS acceleration and jerk in the longitudinal, lateral, and combined directions. Although lane changing increases lateral motion, its substantially lower longitudinal demands result in the smallest combined RMS values. Each row shares the same vertical scale.

Driving strategy	$a_{wx}$	$a_{wy}$	$a_{v,h}$
Reactive driving	1.51	0.17	1.52
Anticipatory braking	1.41	0.16	1.42
Lateral avoidance	0.48	0.23	0.53
ISO 2631-1 Annex C discomfort ranges			RMS range
Not uncomfortable			$<0.315$
A little uncomfortable			0.315–0.63

Table 2: ISO 2631-1 horizontal frequency-weighted RMS acceleration during the three seconds preceding the reactive collision, alongside the approximate discomfort ranges from Annex C of the standard [ 10 ] . All acceleration values are expressed in m/s 2 .

Autonomous Driving Planning

Autonomous driving planning selects safe, feasible vehicle actions and trajectories amid traffic, uncertainty, road rules, and changing goals.

See all 385 papers

Evolutionary Physics-Informed Temporal Fusion for Lane-Change Intention Prediction
Dec 30, 2025Jiazhao Shi, Qiyang Xie, Ziyu Wang +7Autonomous Driving Physics-Informed ML
Tandon School of Engineering New York University New York, USA · Khoury College of Computer Science Northeastern University Boston, USA · School of Business Wake Forest University Winston Salem, USA +6
A Dynamic Scene Interaction Reasoning Framework for Scene-level Lane-Change Intention and Trajectory Prediction of Multiple Interacting Vehicles
Jul 2, 2026Joshua Kofi Asamoah, Blessing Agyei Kyem, Eugene Denteh +1Temporal GNNs Multi-Agent Trajectory Prediction
North Dakota State University, 1410 14th Avenue Offerdahl North Building, CIE 201, Fargo, 58102, North Dakota, United States
From Cues to Horizons: Dynamic Risk Horizon Profiling for Trajectory Prediction
May 30, 2026Xinyi Ning, Zilin Bian, Dachuan Zuo +2Autonomous Driving Risk-Aware Path Planning
Department of Civil and Urban Engineering, New York University. · Department of Civil Engineering Technology and Environmental Management Safety, Rochester Institute of Technology.

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

Evolutionary Physics-Informed Temporal Fusion for Lane-Change Intention Prediction

Jiazhao Shi, Qiyang Xie, Ziyu Wang, Yichen Lin+6

Tandon School of Engineering New York University New York, USA · Khoury College of Computer Science Northeastern University Boston, USA · School of Business Wake Forest University Winston Salem, USA+6

Autonomous Driving Physics-Informed ML

Open paper AskPDF arXiv

Early lane-change intention prediction is essential for autonomous driving and ADAS, but it remains challenging because lane-changing behavior depends on evolving traffic risk, surrounding-vehicle interactions, and target-lane feasibility rather than only instantaneous vehicle states. This study proposes an evolutionary physics-informed temporal fusion framework for three-class lane-change intention prediction, including left lane change, right lane change, and no lane change. Instead of using static physics-informed variables alone, the proposed method derives temporal descriptors from conventional traffic signals, including risk evolution, gap persistence, counterfactual lane utility, interaction pressure gradient, maneuver feasibility, and intent consistency. These descriptors are fused with temporal embeddings learned from raw trajectory sequences through a sequence encoder, and the fused representation is used for final classification. Experiments are conducted on the highD and exiD datasets under 1,s, 2,s, and 3,s prediction horizons. The proposed model achieves Macro F1-scores of 0.9514, 0.9256, and 0.8872 on highD, and 0.9386, 0.9070, and 0.8531 on exiD, respectively. The improvement is especially pronounced in exiD ramp-adjacent scenarios, indicating that temporal physical evolution is particularly useful in interaction-rich environments. These results demonstrate that combining evolutionary physics-informed descriptors with learned temporal representations provides a more dynamic and interpretable solution for early lane-change intention prediction.

Abstract

Autonomous Driving Planning

Explore similar work

Evolutionary Physics-Informed Temporal Fusion for Lane-Change Intention Prediction

A Dynamic Scene Interaction Reasoning Framework for Scene-level Lane-Change Intention and Trajectory Prediction of Multiple Interacting Vehicles

From Cues to Horizons: Dynamic Risk Horizon Profiling for Trajectory Prediction