ROSA: Roundabout Optimized Speed Advisory with Multi-Agent Trajectory Prediction in Multimodal Traffic

Anna-Lena Schlamp; Jeremias Gerner; Klaus Bogenberger; Werner Huber; Stefanie Schmidtner

cs.MAFeb 16, 2026

ROSA: Roundabout Optimized Speed Advisory with Multi-Agent Trajectory Prediction in Multimodal Traffic

Authors: Anna-Lena Schlamp, Jeremias Gerner, Klaus Bogenberger, Werner Huber, Stefanie Schmidtner

Organizations: Institute AImotion Bavaria, Technische Hochschule Ingolstadt, Ingolstadt, Germany · School of Engineering and Design, Technical University of Munich, Munich, Germany · CARISSMA Institute of Automated Driving (C-IAD), Technische Hochschule Ingolstadt, Ingolstadt, Germany

Abstract

We present ROSA -- Roundabout Optimized Speed Advisory -- a system that combines multi-agent trajectory prediction with coordinated speed guidance for multimodal, mixed traffic at roundabouts. Using a Transformer-based model, ROSA jointly predicts the future trajectories of vehicles and Vulnerable Road Users (VRUs) at roundabouts. Trained for single-step prediction and deployed autoregressively, it generates deterministic outputs, enabling actionable speed advisories. Incorporating motion dynamics, the model achieves high accuracy (ADE: 1.29m, FDE: 2.99m at a five-second prediction horizon), surpassing prior work. Adding route intention further improves performance (ADE: 1.10m, FDE: 2.36m), demonstrating the value of connected vehicle data. Based on predicted conflicts with VRUs and circulating vehicles, ROSA provides real-time, proactive speed advisories for approaching and entering the roundabout. Despite prediction uncertainty, ROSA significantly improves vehicle efficiency and safety, with positive effects even on perceived safety from a VRU perspective. The source code of this work is available under: github.com/urbanAIthi/ROSA.

Figures & tables

Paper	Model	Roundabout	Vehicles	VRUs	Multi-Agent	Agent Dynamics	History Length	Prediction Horizon	ADE / FDE
Vanilla-TF [ 10 ]	TF	✓	✓	–	–	✓	–	5.0	1.88 / 4.85
IAMP [ 11 ]	Dynamic Bayesian Network	✓	✓	–	–	✓	4.0	4.0	1.43 / 3.75
3D-A-W [ 12 ]	Anchor Trajectories + LSTM	✓	✓	–	–	✓	2.0	4.0	– / 3.08
GCN [ 13 ]	Gated Recurrent Unit	✓	✓	–	–	✓	1.0	1.0	0.16 / 0.31
ITRA [ 14 ]	RNN	✓	✓	–	✓	✓	1.0	3.0	0.17 / 0.49
MTP-GO [ 15 ]	Temporal GNN	✓	✓	–	✓	✓	3.0	5.0	0.97 / 3.02

TABLE I : Comparison of related prediction models by design aspects and performance.

Fig. 1 : Visualization of the urban roundabout rdb1 from the openDD dataset [ 31 ] with prioritized VRU crossings. The sample scenario illustrates interaction-aware trajectory prediction by ROSA jointly for all agents, including a cyclist (magenta). Shaded areas highlight a representative crosswalk and entry conflict zone for occupancy evaluation.

Input Features	ADE / FDE
Input Features	1	2	3	4	5
Position and Type (1)	0.70 / 0.70	1.54 / 2.37	2.54 / 4.55	3.73 / 7.28	5.08 / 10.51
+ Dynamics (2)	0.14 / 0.14	0.30 / 0.46	0.54 / 1.01	0.86 / 1.84	1.29 / 2.99
+ Exit Intention (3)	0.14 / 0.14	0.30 / 0.46	0.52 / 0.96	0.79 / 1.59	1.10 / 2.36

TABLE II : Comparison of trajectory prediction accuracy for different input feature configurations. Average Displacement Error (ADE) and Final Displacement Error (FDE) (in meters) are reported over prediction horizons ranging from 1 to 5 seconds.

Prediction Horizon	VRU Crosswalks		Roundabout Entries
	Precision	Recall	Precision	Recall
1	0.97	0.99	0.98	0.98
	0.98	0.98	0.98	0.98
2	0.95	0.94	0.96	0.93
	0.95	0.93	0.96	0.92
3	0.91	0.88	0.85	0.80

TABLE III : Occupancy estimation performance for crosswalk and entry zones at prediction horizons ranging from 1 to 5 seconds. Results for model (2) (motion dynamics) are shown in the top row, and for model (3) (including exit intention) in the bottom row.

Performance Metric	ROSA with Prediction Uncertainty			ROSA with Perfect Foresight
Performance Metric	Optimizable Scenarios	Non-Optimizable Scenarios	All Scenarios	Optimizable Scenarios	Non-Optimizable Scenarios	All Scenarios
BEV Energy Consumption	-9.24%	+0.31%	-1.21%	-16.56%	+/-0.00%	-2.63%
BEV Energy Consumption	-10.16%	+0.10%	-1.53%
Fuel Consumption & CO 2 Emissions	-4.14%	+1.23%	+0.38%	-7.75%	+/-0.00%	-1.22%
Fuel Consumption & CO 2 Emissions	-4.71%	+0.63%	-0.22%
Travel Time	-2.94%	+0.43%	-0.1%	-5.28%	+/-0.00%	-0.84%

TABLE IV : Efficiency and safety improvements of ROSA in 6600 real-world scenarios, averaged across optimizable, non-optimizable, and all scenarios. Results under prediction uncertainty (model (2) in the top row, model (3) in the bottom row) are shown on the left, with perfect foresight presented on the right.

Intelligent Transportation Systems

Intelligent Transportation Systems apply sensing, communication, AI, and optimization to make transport networks safer and more efficient.

See all 301 papers

ROSA-RL: Uncertainty-Aware Roundabout Optimized Speed Advisory with Reinforcement Learning
Jun 15, 2026Anna-Lena Schlamp, Jeremias Gerner, Klaus Bogenberger +2Autonomous Driving Planning RL for Autonomous Driving
Institute AImotion Bavaria Technische Hochschule Ingolstadt Ingolstadt, Germany · School of Engineering and Design Technical University of Munich Munich, Germany · CARISSMA Institute of Automated Driving (C-IAD) Technische Hochschule Ingolstadt Ingolstadt, Germany
RankFormer: A Propose-then-Select Transformer for Multi-Agent Multimodal Trajectory Prediction
Apr 8, 2026Diyi Liu, Zihan Niu, Tu Xu +2Multi-Agent Trajectory Prediction Multimodal Trajectory Prediction
College of Metropolitan Transportation, Beijing University of Technology, Beijing, China · Zhejiang Police College, Zhejiang, China
Collaborative Trajectory Prediction via Late Fusion
Apr 24, 2026Nadya Abdel Madjid, Murad Mebrahtu, Zakhar Yagudin +5Late Fusion Multi-Agent Trajectory Prediction
Khalifa University, Computer Science, Abu Dhabi, UAE · Khalifa University, KUCARS-KU Center for Autonomous Robotic Systems, Department of Computer Science, Abu Dhabi, UAE · Khalifa University, Computer and Information Engineering, Abu Dhabi, UAE +2

Jun 15, 2026J/K move · Enter open · S save

ROSA-RL: Uncertainty-Aware Roundabout Optimized Speed Advisory with Reinforcement Learning

Anna-Lena Schlamp, Jeremias Gerner, Klaus Bogenberger, Werner Huber+1

Institute AImotion Bavaria Technische Hochschule Ingolstadt Ingolstadt, Germany · School of Engineering and Design Technical University of Munich Munich, Germany · CARISSMA Institute of Automated Driving (C-IAD) Technische Hochschule Ingolstadt Ingolstadt, Germany

Autonomous Driving Planning RL for Autonomous Driving

Open paper AskPDF arXiv

Roundabouts challenge automated driving in mixed traffic, as heterogeneous and non-deterministic human behavior, unknown driving intentions, and high interaction complexity create uncertainty about whether the conflict zone will be blocked or available at the moment of entry. We present ROSA-RL -- uncertainty-aware Roundabout Optimized Speed Advisory with Reinforcement Learning. It enables safe and efficient roundabout entry for automated and human-driven vehicles in mixed traffic through probabilistic conflict forecasting. A Transformer-based model predicts conflict zone occupancy over a five-second horizon, capturing multi-agent interactions to anticipate upcoming conflicts and available gaps. The prediction outputs encode uncertainty in future motion and intent, and augment the state of a classical RL framework, enabling uncertainty-aware speed coordination. Evaluated in simulations grounded in real-world data, ROSA-RL can effectively handle uncertainty and outperform a comparable model-based baseline, closing the gap to an ideal setting assuming fully known occupancy while improving traffic efficiency and safety. The source code of this work is available under: github.com/urbanAIthi/ROSA-RL.

Abstract

Intelligent Transportation Systems

Explore similar work

ROSA-RL: Uncertainty-Aware Roundabout Optimized Speed Advisory with Reinforcement Learning

RankFormer: A Propose-then-Select Transformer for Multi-Agent Multimodal Trajectory Prediction

Collaborative Trajectory Prediction via Late Fusion