arXiv:2607.04770v1 [cs.SE] 6 Jul 2026
Cam2Sim: Neural Scenario Reconstruction for Closed-Loop Autonomous Driving Simulation Davide Jannussi
Stefano Carlo Lambertenghi
Politecnico di Torino Torino, Italy [email protected]
TUM, fortiss Munich, Germany [email protected]
Constantin Carste
Andrea Stocco
TUM Munich, Germany [email protected]
TUM, fortiss Munich, Germany [email protected]
Abstract
1
Simulation-based testing enables safe and repeatable evaluation of autonomous driving systems, but its effectiveness is limited by the gap between synthetic simulator outputs and real-world camera observations. To address this problem, we present Cam2Sim, a tool that transforms real-world driving recordings into playable CARLA simulation scenarios. Starting from camera images and poses, Cam2Sim reconstructs road geometry, ego trajectories, parked vehicles, and simulation assets, and augments the reconstructed environment with Gaussian Splatting to render camera observations that resemble the original recording. The framework supports ROSbased data extraction, parked-vehicle detection, OpenStreetMapbased map generation, CARLA scenario construction, Gaussian Splatting training, trajectory replay, and closed-loop execution with a system under test. We validate Cam2Sim on a real-world urban-driving scenario with a camera-based end-to-end driving model, comparing reconstruction quality, image-generation quality, and closed-loop behavior against both a simulation-only baseline and the real-world target. Results show that Gaussian-Splattingbased rendering reduces the visual gap with respect to standard simulator rendering and improves behavioral similarity to the realworld reference runs. The artifact is publicly available at https: //github.com/ast-fortiss-tum/cam2sim, and a screencast showing the tool is available at https://youtu.be/KmZ74l1__lI.
Autonomous driving systems (ADS) must be evaluated before deployment because failures can have safety-critical consequences [26]. Modern ADS rely heavily on machine learning and deep learning components for perception and decision-making [17], whose datadriven nature introduces failure modes that differ fundamentally from those of traditional software, motivating specialized testing techniques [10]. Real-world testing provides the strongest evidence of system performance under realistic operating conditions, but it is expensive, difficult to reproduce, time-consuming, and unsafe for many corner cases [20]. Dataset-based evaluation improves reproducibility, yet it replays fixed observations and therefore cannot capture the closed-loop interaction between the ADS and its environment [21]. As a result, failures emerging from the interaction between perception, planning, and control may remain undetected. Simulation-based testing addresses this limitation by enabling the ADS to interact with a controllable and reproducible virtual environment. However, its effectiveness is fundamentally limited by the sim-to-real gap [20]. Differences in textures, illumination, scene appearance, and other modeling inaccuracies can alter the input distribution perceived by the ADS, leading to behaviors that deviate from those observed in the real world [12, 13]. Consequently, improving the realism and trustworthiness of simulation environments has become a central objective of recent research in ADS testing [2, 19]. Recent advances in neural rendering and scene reconstruction, including GANs, diffusion models, Neural Radiance Fields, and Gaussian Splatting, have shown promise in generating visually realistic scenes from real-world recordings [9, 13, 27]. However, existing approaches such as DriveRecon [15] mainly focus on rendering viewpoints or image sequences rather than executable, closed-loop simulations [1]. Conversely, prior work on scenario reconstruction from police reports and structured accident descriptions focuses on rebuilding executable scenarios [8], but does not address the perception-level sim-to-real gap. To address this gap, we present Cam2Sim, the first framework, to the best of our knowledge, that transforms real-world 2D driving recordings and camera datasets into 3D playable CARLA scenarios for closed-loop ADS testing. Starting from front-facing camera images and poses, Cam2Sim reconstructs road geometry, ego trajectories, parked vehicles, and simulation assets inside CARLA [6].
ACM Reference Format: Davide Jannussi, Stefano Carlo Lambertenghi, Constantin Carste, and Andrea Stocco. 2026. Cam2Sim: Neural Scenario Reconstruction for ClosedLoop Autonomous Driving Simulation. In Proceedings of Proceedings of the 41st IEEE/ACM International Conference on Automated Software Engineering (ASE ’26). ACM, New York, NY, USA, 5 pages. https://doi.org/XXXXXXX. XXXXXXX
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]. ASE ’26, Munich, Germany © 2026 Copyright held by the owner/author(s). Publication rights licensed to ACM. ACM ISBN 978-1-4503-XXXX-X/2026/06 https://doi.org/XXXXXXX.XXXXXXX
Introduction
ASE ’26, October 12–16, 2026, Munich, Germany
The framework then augments the reconstructed environment with Gaussian Splatting [11] to render camera observations that resemble the original recording. The resulting scenarios support both trajectory replay and closed-loop ADS testing, where the ADS receives the GS-rendered camera observations during execution. Cam2Sim supports ROS-based extraction, parked-vehicle detection from camera or LiDAR data, OpenStreetMap-based map generation, CARLA scenario construction, local GS training, trajectory replay, and closed-loop execution with a camera-based ADS. It outputs CARLA-ready maps, trajectories, parked-vehicle placements, GS models, replay videos, steering logs, and execution traces for downstream analysis. We evaluate Cam2Sim on a real-world urban-driving scenario using a camera-based end-to-end driving model. Our evaluation assesses reconstruction quality and closed-loop behavioral fidelity under simulator-rendered and GS-rendered observations. Results show that GS-rendered observations reduce the visual discrepancy with respect to standard simulator rendering while enabling behaviors that more closely resemble the real-world reference runs. This tool paper makes the following contributions: Tool. We present Cam2Sim, a framework that transforms realworld recordings into executable CARLA scenarios for closed-loop ADS testing through reconstruction and Gaussian-Splatting-based rendering in CARLA. Evaluation in a real-world urban driving scenario. We demonstrate improvements in reconstruction quality and behavioral similarity compared to a simulation-only baseline. In our experiments, the GS-augmented simulation successfully completes all runs, while the simulation-only baseline consistently fails the scenario.
2
Cam2Sim Framework
Figure 1 shows the Cam2Sim workflow, which consists of five components: (1) data extraction, (2) dataset processing, (3) simulationasset generation, (4) Gaussian Splatting preparation, and (5) driving simulation. Together, these components transform raw recordings or prepared camera datasets into CARLA-ready scenarios with map data, trajectories, parked-vehicle positions, GS-training inputs, and executable replay or closed-loop simulations.
2.1
Davide Jannussi, Stefano Carlo Lambertenghi, Constantin Carste, and Andrea Stocco
2.2
Dataset Processing
The dataset processing component extracts the information required for scenario reconstruction, map generation, GS training, and simulation. The camera trajectory is used as the reference path for reconstructing and replaying the scenario. Cam2Sim detects parked vehicles and provides a graphical interface for manual refinement. From camera data, it applies monocular 3D object detection with FCOS3D [24], transforms detections into world coordinates using camera-to-ego calibration and timestampaligned ego poses, and clusters repeated detections across frames. Final positions are computed as confidence-weighted averages [22], while lane-side and orientation labels are obtained through majority voting. When LiDAR is available, Cam2Sim can instead detect parked vehicles with PointPillars [14]; users can then move, rotate, insert, or delete boxes to produce refined parked-vehicle files. This component also prepares map and GS inputs. From the trajectory or a manually specified address, Cam2Sim retrieves OpenStreetMap data through the Overpass API [7]. For GS training, it selects frames at a configurable interval, removes the visible egovehicle hood from the selected frames, generates sky masks using SegFormer trained on Cityscapes [5, 25], and splits the route into overlapping chunks containing images, masks, and pose metadata.
2.3
Simulation Data Generation
The simulation-data generation component converts the processed scenario into CARLA-ready assets: the OpenDRIVE map, replay trajectory, initial hero-vehicle pose, and parked-vehicle spawn positions. Cam2Sim exports both center-position and rear-axle trajectory variants to account for the difference between recorded sensor poses and CARLA vehicle transforms. Coordinate conversion maps dataset positions to CARLA coordinates through the OpenDRIVE map. When geographic coordinates are used, Cam2Sim converts WGS84 coordinates into the OpenDRIVE reference frame and adapts the axis convention for CARLA. The tool also provides utilities to load the generated map, inspect trajectory alignment and parked-vehicle placement, and prepare the final CARLA scenario.
Data Extraction
The data extraction component converts ROS-based recordings into the structured dataset format used by Cam2Sim. It can be skipped when front-facing camera images and corresponding camera poses are already available. The minimum required input is a sequence of camera images and poses, which define the camera trajectory and are sufficient for scenario reconstruction, GS training, and replay. When ROS bags are used, Cam2Sim extracts camera frames and associates each frame with the ego pose at the corresponding timestamp. It can also extract optional LiDAR, odometry, steering angles, and model-prediction streams. These signals improve reconstruction and support validation: LiDAR enables more accurate parked-vehicle detection, while steering and model-prediction logs support comparison with the ADS. Sensor streams with different frequencies are synchronized with ego poses through timestampbased interpolation. The output is a structured dataset containing camera frames, poses, and point clouds, and trajectories.
2.4
Gaussian Splatting Preparation
Using the processed frames and reconstructed trajectories, the GS preparation component trains the models used to render camera observations that resemble the original recording. Cam2Sim trains local GS models to improve scalability and specialization to local scene appearance. Cam2Sim runs COLMAP [18] with the selected images, calibrated camera model, and optional sky masks to estimate camera poses and a sparse 3D point cloud. It then trains GS models with Nerfstudio [23], using splatfacto or splatfacto-big; sky masks exclude sky regions so the model focuses on static scene content. After training, Cam2Sim aligns the Nerfstudio and dataset coordinate systems by matching reconstructed camera poses with the original pose annotations and estimating the corresponding alignment parameters. These models are later used to render GS images from CARLA camera poses.
Run
Speed Camera PID Control
E2E Model
Throttle # "Steer
Ultra7 $ " 64 GB RTX5080
"
Cam2Sim : Neural Scenario Reconstruction for Closed-Loop Autonomous Driving Simulation
C B A
1
• Trajectory • Parked
ASE ’26, October 12–16, 2026, Munich, Germany
3
"#
" SuT
vehicles
! Camera
! Localization ! LiDAR
2
• Map • Images
4
5
Figure 1: Overview of Cam2Sim: (1) Data Extraction, (2) Dataset Processing, (3) Simulation Data Generation, (4) Gaussian Splatting Preparation, and (5) Driving Simulation. Because the deployed ADS directly consumes the rendered camera stream, successful closed-loop execution also provides an operational assessment of the generated observations: severe rendering artifacts or missing scene information would directly affect steering behavior and trajectory stability.
2.5
Driving Simulation
The driving simulation component executes the reconstructed scenario in CARLA using the generated map, trajectories, parked vehicles, camera configuration, and, when enabled, the trained GS models and alignment files. Cam2Sim supports both trajectory replay and closed-loop execution. In trajectory-replay mode, the hero vehicle follows the recorded trajectory frame by frame. This mode is used to inspect the reconstructed scenario and compare observations from the same viewpoints as the original recording. Cam2Sim can save CARLArendered images, GS-rendered images, and side-by-side outputs. In closed-loop mode, the ADS controls the ego vehicle. Cam2Sim currently supports a camera-based DAVE-2 model [3], served by a separate inference process that predicts steering commands from RGB images. The ADS can receive either raw CARLA images, forming the simulation-only baseline, or GS-rendered images to evaluate whether GS-based rendering reduces the gap to the original driving scenario. The output includes replayed camera streams, trajectories, steering commands, and execution logs for evaluating reconstruction quality, visual fidelity, and behavior similarity.
3
Preliminary Validation
We evaluate whether Cam2Sim can transform a real-world driving recording into a reconstructed simulation that is geometrically consistent with the original scenario, visually closer to the original camera stream, and suitable for closed-loop ADS testing. The validation follows the structure of the pipeline. First, we assess whether the reconstruction and simulation-generation components correctly reproduce the road geometry, ego trajectory, and parked vehicles of the target scenario. Second, we evaluate whether Gaussian Splatting preserves visually relevant scene information required for closed-loop ADS execution. To this aim, we conduct a real-world data-collection campaign using an ADS-equipped research vehicle on a public urban road and use the resulting recording as the target scenario for Cam2Sim. The evaluation compares three domains: (i) the real-world reference execution, (ii) a simulation-only baseline using the reconstructed
CARLA scenario, and (iii) a GS-augmented simulation in which camera observations are rendered using Gaussian Splatting. System Under Test. We use a camera-based end-to-end DAVE2 model [3] that predicts continuous steering directly from RGB images captured by the front-facing camera. During execution, longitudinal control is handled independently by a PID controller that maintains a fixed speed of 20 km/h (Figure 2 (A)). Autonomous Driving Platform. We use fortuna [4], a modified Volkswagen Passat Variant GTE research vehicle equipped with cameras, Velodyne LiDARs, and GNSS/INS localization. The frontfacing camera is used both for scenario reconstruction and as input to the ADS, while the roof-mounted LiDAR supports parked-vehicle detection during real-world data collection (Figure 2 (B)). Collection Scenario. Experiments are conducted in Munich on a 450 m two-way residential street with a 30 km/h speed limit. The scenario includes an unmarked roadway, sidewalks, three bends, and parked vehicles partially narrowing the driving lane, providing a realistic yet controlled urban evaluation environment (Figure 2 (C)).
3.1
Scenario Reconstruction Quality
We first evaluate whether Cam2Sim reconstructs the target scenario with sufficient geometric fidelity for simulation-based testing. We apply the reconstruction pipeline to the collected recording and generate a CARLA scenario containing the reconstructed road map, replay trajectory, and parked-vehicle placements. We then execute trajectory replay and compare the rendered semantic maps against the real-world semantic observations from the same viewpoints. We assess reconstruction quality using Intersection over Union (IoU) over the road, car, and background classes. Across 3,144 replayed frames, Cam2Sim achieves a mean IoU (mIoU) of 0.774 (𝜎 = 0.071). The reconstructed road layout achieves the highest agreement with the real-world reference, with a mean IoU of 0.847 (𝜎 = 0.056), while background regions achieve 0.889 (𝜎 = 0.037). Vehicle reconstruction is more challenging due to localization inaccuracies and partial occlusions, yet the car class still achieves a mean IoU of 0.577 (𝜎 = 0.206). An example semantic comparison is shown in Figure 2 (D). Overall, these results suggest that Cam2Sim reconstructs urban-driving scenarios with sufficient geometric fidelity to support replay and closed-loop ADS testing.
3.2
Image Generation Quality
We next evaluate whether GS reduces the visual gap between the reconstructed CARLA scenario and the real recording. Starting from the reconstructed scenario, we train the GS model using Step 4 on
ASE ’26, October 12–16, 2026, Munich, Germany
A
%#
B #
Davide Jannussi, Stefano Carlo Lambertenghi, Constantin Carste, and Andrea Stocco
!C
Speed Camera PID Control
E2E Model
Throttle" !Steer
E
D
SIM
GS
RW
" Ultra7 ! " 64 GB
SIM
RTX5080
Figure 2: Validation setup and results: (A) the ADS, (B) the autonomous-driving platform used to collect the target recording, (C) the real-world collection scenario, (D) results of Scenario Reconstruction, and (E) Behavior Fidelity.
C
• Trajectory
the prepared camera frames, poses, sky masks, and local training • Parked splits. We then execute Step 5 again in trajectory-replay mode and vehicles B stream with GS from the replayed viewpoints. render the camera Map • ! Camera We compare three image streams: the real-world camera frames, Localization !CARLA the frames, and2 the GS-rendered frames. • Images A simulation-only 1 !whether LiDARGS produces observations closer to The goal is to verify the real camera stream than the raw simulator output. We quantify image quality using three full-reference metrics: TO DO: : ▶METRIC 1◀ , TO DO: : ▶METRIC 2◀ , and TO DO: : ▶METRIC 3◀ . Results are shown in TO DO: : ▶XXX ◀ . Andrea: ▶complete◀
3
" SuT "# Table 1: Behavior fidelity comparison between the realworld execution, the simulation-only baseline, and the GSaugmented simulation. Higher completion rates and lower trajectory-deviation metrics 5 indicate behavior closer to the real-world reference. OR = out of road, CC = car crash. 4 Metric Failure Rate Completion Rate (%) Failure Type (OR–CC)
Real
Cam2Sim
Sim
0/3 100–100–100 –
0/3 100–100–100 –
3/3 21–23–20 2–1
Trajectory Similarity
3.3
Finally, we evaluate whether the GS-augmented simulation improves system-level testing realism. We execute the driving-simulation component in closed-loop mode with the deployed ADS in three domains: the real-world setting, a simulation-only baseline, and a GS-augmented simulation. In the simulation-only baseline, the ADS receives raw CARLA camera images; in the GS-augmented setting, it receives camera observations rendered by the GS model. To account for variability and the non-determinism of the endto-end driving model, we execute each domain three times. We compare the generated trajectories against the real-world reference executions using failure and completion rates, as well as trajectorysimilarity and steering-smoothness metrics, including Fréchet distance, corridor violations, lateral excess, and steering jitter. Results are shown in Figure 2 (E) and Table 1. The simulationonly baseline consistently fails to complete the scenario, while the GS-augmented simulation successfully completes all runs, matching the real-world executions. Moreover, the GS-based simulation produces trajectories substantially closer to the real-world reference and produces stable steering behavior across all runs. Overall, these preliminary results suggest that GS-rendered observations improve the robustness and behavioral fidelity of closed-loop ADS testing compared to standard simulator rendering.
4
Fréchet Distance (m) Corridor Violations (%) Mean Excess (m) Excess When Out (m)
Behavior Fidelity
Conclusions
We presented Cam2Sim, a tool that transforms real-world recordings or prepared camera datasets into playable CARLA scenarios augmented with Gaussian Splatting. Starting from camera images and poses, the tool reconstructs scenario geometry, prepares simulation-ready assets, trains local GS models, and supports both trajectory replay and ADS closed-loop execution. As future work, we plan to extend Cam2Sim with dynamic-object reconstruction, support for additional neural-rendering backends
0.00 0.0 0.00 0.00
2.22 67.3 0.330 0.491
– – – –
0.867 4.983
– –
Steering Smoothness Average Jitter (rad/s) Maximum Jitter (rad/s)
1.167 6.452
and sensor modalities, and larger-scale evaluations across diverse driving environments and ADS architectures.
5
Data Availability Statement
Cam2Sim is released as a GitHub repository under the Apache-2.0 license [16]. The artifact includes the tool implementation, the ROS bag used in our validation, the camera-based ADS, example outputs, and scripts for reproducing the GS-rendering and closed-loop evaluations. The repository also documents the required dependencies, including CARLA, ROS, COLMAP, Nerfstudio, and a CUDA-capable GPU. Our validation setup uses an Intel Core Ultra 9 machine with 32 GB RAM and an NVIDIA RTX 4090 GPU. The review dataset is provided for reproducibility purposes only; an anonymized version with blurred faces and license plates will be publicly released under a CC BY-NC 4.0 license. A public DOI is not released at submission time to avoid premature dissemination; the artifact is available as a GitHub repository for review and will be archived with a DOI upon acceptance. Andrea:
▶revise◀
References [1] Luciano Baresi, Davide Yi Xian Hu, Andrea Stocco, and Paolo Tonella. 2025. Efficient Domain Augmentation for Autonomous Driving Testing Using Diffusion Models. In Proceedings of 47th International Conference on Software Engineering (ICSE ’25). IEEE. [2] Matteo Biagiola, Andrea Stocco, Vincenzo Riccio, and Paolo Tonella. 2024. Two is better than one: digital siblings to improve autonomous driving testing. Empirical
Cam2Sim : Neural Scenario Reconstruction for Closed-Loop Autonomous Driving Simulation
Softw. Engg. 29, 4 (May 2024), 33 pages. doi:10.1007/s10664-024-10458-4 [3] Mariusz Bojarski, Davide Del Testa, Daniel Dworakowski, Bernhard Firner, Beat Flepp, Prasoon Goyal, Lawrence D. Jackel, Mathew Monfort, Urs Muller, Jiakai Zhang, Xin Zhang, Jake Zhao, and Karol Zieba. 2016. End to End Learning for Self-Driving Cars. arXiv:1604.07316 [cs.CV] [4] Martin Buechel, Malte Schellmann, Holger Rosier, Tobias Kessler, and Alois Knoll. 2019. Fortuna: Presenting the 5G-connected automated vehicle prototype of the project PROVIDENTIA. Researchgate. https://doi. org/10.13140/RG 2, 24402.91842 (2019). doi:10.13140/rg.2.2.24402.91842 [5] Marius Cordts, Mohamed Omran, Sebastian Ramos, Timo Rehfeld, Markus Enzweiler, Rodrigo Benenson, Uwe Franke, Stefan Roth, and Bernt Schiele. 2016. The Cityscapes Dataset for Semantic Urban Scene Understanding. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 3213–3223. doi:10.1109/CVPR.2016.350 [6] Alexey Dosovitskiy, German Ros, Felipe Codevilla, Antonio Lopez, and Vladlen Koltun. 2017. CARLA: An Open Urban Driving Simulator. In Proceedings of the 1st Annual Conference on Robot Learning (Proceedings of Machine Learning Research, Vol. 78), Sergey Levine, Vincent Vanhoucke, and Ken Goldberg (Eds.). PMLR, 1–16. https://proceedings.mlr.press/v78/dosovitskiy17a.html [7] Overpass-API Drolbr. 2024. Drolbr/overpass-API: A database engine to query the OpenStreetMap data. [8] Alessio Gambi, Tri Huynh, and Gordon Fraser. 2019. Automatically Reconstructing Car Crashes from Police Reports for Testing Self-Driving Cars. In 2019 IEEE/ACM 41st International Conference on Software Engineering: Companion Proceedings (ICSE-Companion). 290–291. doi:10.1109/ICSE-Companion.2019.00119 [9] Yuan Gao, Mattia Piccinini, Yuchen Zhang, Dingrui Wang, Korbinian Moller, Roberto Brusnicki, Baha Zarrouki, Alessio Gambi, Jan Frederik Totz, Kai Storms, Steven Peters, Andrea Stocco, Bassam Alrifaee, Marco Pavone, and Johannes Betz. 2026. Foundation Models in Autonomous Driving: A Survey on Scenario Generation and Scenario Analysis. IEEE Open Journal of Intelligent Transportation Systems (2026), 1–1. doi:10.1109/OJITS.2026.3660686 [10] Nargiz Humbatova, Gunel Jahangirova, Gabriele Bavota, Vincenzo Riccio, Andrea Stocco, and Paolo Tonella. 2020. Taxonomy of real faults in deep learning systems. In Proceedings of the ACM/IEEE 42nd international conference on software engineering. doi:10.1145/3377811.3380395 [11] Bernhard Kerbl, Georgios Kopanas, Thomas Leimkuehler, and George Drettakis. 2023. 3D Gaussian Splatting for Real-Time Radiance Field Rendering. ACM Trans. Graph. 42, 4, Article 139 (July 2023), 14 pages. doi:10.1145/3592433 [12] Stefano Carlo Lambertenghi, Mirena Flores Valdez, and Andrea Stocco. 2025. A Multi-Modality Evaluation of the Reality Gap in Autonomous Driving Systems. In 2025 40th IEEE/ACM International Conference on Automated Software Engineering (ASE). 2808–2820. doi:10.1109/ASE63991.2025.00230 [13] Stefano Carlo Lambertenghi and Andrea Stocco. 2024. Assessing Quality Metrics for Neural Reality Gap Input Mitigation in Autonomous Driving Testing. In 2024 IEEE Conference on Software Testing, Verification and Validation (ICST). IEEE Computer Society, Los Alamitos, CA, USA, 173–184. doi:10.1109/ICST60714.2024. 00024 [14] Alex H. Lang, Sourabh Vora, Holger Caesar, Lubing Zhou, Jiong Yang, and Oscar Beijbom. 2019. PointPillars: Fast Encoders for Object Detection From Point Clouds. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition
ASE ’26, October 12–16, 2026, Munich, Germany
(CVPR). 12689–12697. doi:10.1109/CVPR.2019.01298 [15] Hao LU, Tianshuo Xu, Wenzhao Zheng, Yunpeng Zhang, Wei Zhan, Dalong Du, Masayoshi Tomizuka, Kurt Keutzer, and Ying-Cong Chen. 2026. DrivingRecon: Large 4D Gaussian Reconstruction Model For Autonomous Driving. In The Thirtyninth Annual Conference on Neural Information Processing Systems. https:// openreview.net/forum?id=HF5A73jmxq [16] replication-package 2026. Replication package. https://github.com/ast-fortisstum/cam2sim. [17] Vincenzo Riccio, Gunel Jahangirova, Andrea Stocco, Nargiz Humbatova, Michael Weiss, and Paolo Tonella. 2020. Testing Machine Learning based Systems: A Systematic Mapping. Empirical Software Engineering (2020). doi:10.1007/s10664020-09881-0 [18] Johannes L. Schönberger and Jan-Michael Frahm. 2016. Structure-from-Motion Revisited. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 4104–4113. doi:10.1109/CVPR.2016.445 [19] Lev Sorokin, Matteo Biagiola, and Andrea Stocco. 2026. Simulator ensembles for trustworthy autonomous driving systems testing. Empirical Softw. Engg. 31, 4 (Feb. 2026), 39 pages. doi:10.1007/s10664-026-10821-7 [20] Andrea Stocco, Brian Pulfer, and Paolo Tonella. 2023. Mind the Gap! A Study on the Transferability of Virtual Versus Physical-World Testing of Autonomous Driving Systems. IEEE Transactions on Software Engineering 49, 04 (apr 2023). doi:10.1109/TSE.2022.3202311 [21] Andrea Stocco, Brian Pulfer, and Paolo Tonella. 2023. Model vs system level testing of autonomous driving systems: a replication and extension study. Empirical Software Engineering 28, 3 (May 2023), 26 pages. doi:10.1007/s10664-023-10306-x [22] Andrea Stocco and Paolo Tonella. 2022. Confidence-driven weighted retraining for predicting safety-critical failures in autonomous driving systems. Journal of Software: Evolution and Process 34, 10 (2022), e2386. doi:10.1002/smr.2386 [23] Matthew Tancik, Ethan Weber, Evonne Ng, Ruilong Li, Brent Yi, Terrance Wang, Alexander Kristoffersen, Jake Austin, Kamyar Salahi, Abhik Ahuja, David Mcallister, Justin Kerr, and Angjoo Kanazawa. 2023. Nerfstudio: A Modular Framework for Neural Radiance Field Development. In ACM SIGGRAPH 2023 Conference Proceedings (Los Angeles, CA, USA) (SIGGRAPH ’23). Association for Computing Machinery, New York, NY, USA, Article 72, 12 pages. doi:10.1145/3588432.3591516 [24] Tai Wang, Xinge Zhu, Jiangmiao Pang, and Dahua Lin. 2021. FCOS3D: Fully Convolutional One-Stage Monocular 3D Object Detection. In 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). 913–922. doi:10. 1109/ICCVW54120.2021.00107 [25] Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, and Ping Luo. 2021. SegFormer: simple and efficient design for semantic segmentation with transformers. In Proceedings of the 35th International Conference on Neural Information Processing Systems (NIPS ’21). Curran Associates Inc., Red Hook, NY, USA, Article 924, 14 pages. [26] Ziyuan Zhong, Yun Tang, Yuan Zhou, Vania de Oliveira Neves, Yang Liu, and Baishakhi Ray. 2021. A Survey on Scenario-Based Testing for Automated Driving Systems in High-Fidelity Simulation. arXiv:2112.00964 [cs.SE] [27] Huixin Zhu, Zhili Zhang, Junyang Zhao, Hui Duan, Yao Ding, Xiongwu Xiao, and Junsong Yuan. 2024. Scene reconstruction techniques for autonomous driving: a review of 3D Gaussian splatting. Artificial Intelligence Review 58 (2024). doi:10. 1007/s10462-024-10955-4