Miguel Angel Munoz-Banon
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Miguel Angel Munoz-Banon has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Fernando Torres
Automation, Robotics, and Computer Vision Group (AUROVA), University of Alicante, Alicante, Spain
9 shared publications - Francisco Andres Candelas
Group of Automation, Robotics and Computer Vision (AUROVA), University of Alicante, Alicante, Spain
8 shared publications - Alejandro Olivas
Automation, Robotics, and Computer Vision Group (AUROVA), University of Alicante, Alicante, Spain
3 shared publications - Edison Velasco-Sánchez
Group of Automation, Robotics and Computer Vision (AUROVA), University of Alicante, 03690 Alicante, Spain
3 shared publications - Iván del Pino
Institut de Robòtica i Informàtica Industrial CSIC-UPC
2 shared publications - Santiago T. Puente
AUROVA Lab, Department of Physics Systems Engineering and Signal Theory, Computer Science Research Institute, University of Alicante, 03690 Alicante, Spain
2 shared publications
Latest Publications
- Open Access
- 18 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 81979-81993
End-to-End Imitation Learning for Socially Aware Navigation Using Memory-Based Trajectory and Action Planning
The increasing deployment of service robots and autonomous vehicles operating in human-populated environments underscores the importance of planning trajectories to ensure pedestrian comfort while optimizing robot travel times. End-to-end models are promising for shortcutting the complex problem of pedestrian detection, trajectory prediction, and robot movement planning by using a single network. Imitation Learning (IL) has been widely adopted in self-driving because it enables implicit understanding of complex environments and expert driving behaviour, whereas socially aware navigation has usually relied on Reinforcement Learning (RL) approaches that learn from abstracted environment representations, prioritizing information relevant to human interaction. In this research, we present Memory-based Trajectory and Action Planning (MTAP), an IL approach extended for socially aware navigation to perform intelligent pedestrian avoidance while maintaining a deeper understanding of the environment and the expert driving behaviour. Our method predicts the trajectory and the control action commands directly from the information of the sensors and the global target, employing a recurrent layer to anticipate pedestrian movements based on past frames. The network is trained and evaluated in the CARLA simulator, where we introduce a benchmark for assessing pedestrian avoidance efficacy, presenting also new social navigation metrics to measure both trajectory efficiency and pedestrian comfort during avoidance maneuvers.
- 18 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 81979-81993
Open AccessEnd-to-End Imitation Learning for Socially Aware Navigation Using Memory-Based Trajectory and Action Planning
The increasing deployment of service robots and autonomous vehicles operating in human-populated environments underscores the importance of planning trajectories to ensure pedestrian comfort while optimizing robot travel times. End-to-end models are promising for shortcutting the complex problem of pedestrian detection, trajectory prediction, and robot movement planning by using a single network. Imitation Learning (IL) has been widely adopted in self-driving because it enables implicit understanding of complex environments and expert driving behaviour, whereas socially aware navigation has usually relied on Reinforcement Learning (RL) approaches that learn from abstracted environment representations, prioritizing information relevant to human interaction. In this research, we present Memory-based Trajectory and Action Planning (MTAP), an IL approach extended for socially aware navigation to perform intelligent pedestrian avoidance while maintaining a deeper understanding of the environment and the expert driving behaviour. Our method predicts the trajectory and the control action commands directly from the information of the sensors and the global target, employing a recurrent layer to anticipate pedestrian movements based on past frames. The network is trained and evaluated in the CARLA simulator, where we introduce a benchmark for assessing pedestrian avoidance efficacy, presenting also new social navigation metrics to measure both trajectory efficiency and pedestrian comfort during avoidance maneuvers.
- Open Access
- 6 March 2026
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 172
Sim-to-real transfer by hybrid Gaussian Splatting and geometric reconstruction for autonomous driving
- 6 March 2026
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 172
Open AccessSim-to-real transfer by hybrid Gaussian Splatting and geometric reconstruction for autonomous driving
- Top Cited PapersOpen Access
- 20 September 2025
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 45 (6) , 839-857
The Oxford Spires Dataset: Benchmarking large-scale LiDAR-visual localisation, reconstruction and radiance field methods
This paper introduces a large-scale multi-modal dataset captured in and around well-known landmarks in Oxford using a custom-built multi-sensor perception unit as well as a millimetre-accurate map from a Terrestrial LiDAR Scanner (TLS). The perception unit includes three synchronised global shutter colour cameras, an automotive 3D LiDAR scanner, and an inertial sensor – all precisely calibrated. We also establish benchmarks for tasks involving localisation, reconstruction, and novel-view synthesis, which enable the evaluation of Simultaneous Localisation and Mapping (SLAM) methods, Structure-from-Motion (SfM) and Multi-view Stereo (MVS) methods as well as radiance field methods such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting. To evaluate 3D reconstruction, the TLS 3D models are used as ground truth. Localisation ground truth is computed by registering the mobile LiDAR scans to the TLS 3D models. Radiance field methods are evaluated not only with poses sampled from the input trajectory, but also from viewpoints that are from trajectories which are distant from the training poses. Our evaluation demonstrates a key limitation of state-of-the-art radiance field methods: we show that they tend to overfit to the training poses/images and do not generalise well to out-of-sequence poses. They also underperform in 3D reconstruction compared to MVS systems using the same visual inputs. Our dataset and benchmarks are intended to facilitate better integration of radiance field methods and SLAM systems. The raw and processed data, along with software for parsing and evaluation, can be accessed at https://dynamic.robots.ox.ac.uk/datasets/oxford-spires/ .
- 20 September 2025
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 45 (6) , 839-857
Top Cited PapersOpen AccessThe Oxford Spires Dataset: Benchmarking large-scale LiDAR-visual localisation, reconstruction and radiance field methods
This paper introduces a large-scale multi-modal dataset captured in and around well-known landmarks in Oxford using a custom-built multi-sensor perception unit as well as a millimetre-accurate map from a Terrestrial LiDAR Scanner (TLS). The perception unit includes three synchronised global shutter colour cameras, an automotive 3D LiDAR scanner, and an inertial sensor – all precisely calibrated. We also establish benchmarks for tasks involving localisation, reconstruction, and novel-view synthesis, which enable the evaluation of Simultaneous Localisation and Mapping (SLAM) methods, Structure-from-Motion (SfM) and Multi-view Stereo (MVS) methods as well as radiance field methods such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting. To evaluate 3D reconstruction, the TLS 3D models are used as ground truth. Localisation ground truth is computed by registering the mobile LiDAR scans to the TLS 3D models. Radiance field methods are evaluated not only with poses sampled from the input trajectory, but also from viewpoints that are from trajectories which are distant from the training poses. Our evaluation demonstrates a key limitation of state-of-the-art radiance field methods: we show that they tend to overfit to the training poses/images and do not generalise well to out-of-sequence poses. They also underperform in 3D reconstruction compared to MVS systems using the same visual inputs. Our dataset and benchmarks are intended to facilitate better integration of radiance field methods and SLAM systems. The raw and processed data, along with software for parsing and evaluation, can be accessed at https://dynamic.robots.ox.ac.uk/datasets/oxford-spires/ .
- Open Access
- 8 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 65428-65443
ViKi-HyCo: A Hybrid-Control Approach for Complex Car-Like Maneuvers
While Visual Servoing is deeply studied to perform simple maneuvers, the literature does not commonly address complex cases where the target is far out of the camera’s field of view (FOV) during the maneuver. For this reason, in this paper, we present ViKi-HyCo (Visual Servoing and Kinematic Hybrid-Controller). This approach generates the necessary maneuvers for the complex positioning of a non-holonomic mobile robot in outdoor environments. In this method, we use LiDAR-camera fusion to estimate objects bounding boxes using image and metrics modalities. With the multi-modality nature of our representation, we can automatically obtain a target for a visual servoing controller. At the same time, we also have a metric target, which allows us to hybridize with a kinematic controller. Given this hybridization, we can perform complex maneuvers even when the target is far away from the camera’s FOV. The proposed approach does not require an object-tracking algorithm and can be applied to any robotic positioning task where its kinematic model is known. ViKi-HyCo has an error of $\mathbf {0.0428\pm 0.0467}$ m in the X-axis and $\mathbf {0.0515\pm 0.0323}$ m in the Y-axis at the end of a complete positioning task.
- 8 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 65428-65443
Open AccessViKi-HyCo: A Hybrid-Control Approach for Complex Car-Like Maneuvers
While Visual Servoing is deeply studied to perform simple maneuvers, the literature does not commonly address complex cases where the target is far out of the camera’s field of view (FOV) during the maneuver. For this reason, in this paper, we present ViKi-HyCo (Visual Servoing and Kinematic Hybrid-Controller). This approach generates the necessary maneuvers for the complex positioning of a non-holonomic mobile robot in outdoor environments. In this method, we use LiDAR-camera fusion to estimate objects bounding boxes using image and metrics modalities. With the multi-modality nature of our representation, we can automatically obtain a target for a visual servoing controller. At the same time, we also have a metric target, which allows us to hybridize with a kinematic controller. Given this hybridization, we can perform complex maneuvers even when the target is far away from the camera’s FOV. The proposed approach does not require an object-tracking algorithm and can be applied to any robotic positioning task where its kinematic model is known. ViKi-HyCo has an error of $\mathbf {0.0428\pm 0.0467}$ m in the X-axis and $\mathbf {0.0515\pm 0.0323}$ m in the Y-axis at the end of a complete positioning task.
- Open Access
- 2 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (6) , 5599-5606
Geo-Localization Based on Dynamically Weighted Factor-Graph
Feature-based geo-localization relies on associating features extracted from aerial imagery with those detected by the vehicle's sensors. This requires that the type of landmarks must be observable from both sources. This lack of variety of feature types generates poor representations that lead to outliers and deviations produced by ambiguities and lack of detections, respectively. To mitigate these drawbacks, in this letter, we present a dynamically weighted factor graph model for the vehicle's trajectory estimation. The weight adjustment in this implementation depends on information quantification in the detections performed using a LiDAR sensor. Also, a prior (GNSS-based) error estimation is included in the model. Then, when the representation becomes ambiguous or sparse, the weights are dynamically adjusted to rely on the corrected prior trajectory, mitigating outliers and deviations in this way. We compare our method against state-of-the-art geo-localization ones in a challenging and ambiguous environment, where we also cause detection losses. We demonstrate mitigation of the mentioned drawbacks where the other methods fail.
- 2 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (6) , 5599-5606
Open AccessGeo-Localization Based on Dynamically Weighted Factor-Graph
Feature-based geo-localization relies on associating features extracted from aerial imagery with those detected by the vehicle's sensors. This requires that the type of landmarks must be observable from both sources. This lack of variety of feature types generates poor representations that lead to outliers and deviations produced by ambiguities and lack of detections, respectively. To mitigate these drawbacks, in this letter, we present a dynamically weighted factor graph model for the vehicle's trajectory estimation. The weight adjustment in this implementation depends on information quantification in the detections performed using a LiDAR sensor. Also, a prior (GNSS-based) error estimation is included in the model. Then, when the representation becomes ambiguous or sparse, the weights are dynamically adjusted to rely on the corrected prior trajectory, mitigating outliers and deviations in this way. We compare our method against state-of-the-art geo-localization ones in a challenging and ambiguous environment, where we also cause detection losses. We demonstrate mitigation of the mentioned drawbacks where the other methods fail.
- 25 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 7 (4) , 12339-12346
Robust Self-Tuning Data Association for Geo-Referencing Using Lane Markings
Localization in aerial imagery-based maps offers many advantages, such as global consistency, geo-referenced maps, and the availability of publicly accessible data. However, the landmarks that can be observed from both aerial imagery and on-board sensors is limited. This leads to ambiguities or aliasing during the data association. Building upon a highly informative representation (that allows efficient data association), this paper presents a complete pipeline for resolving these ambiguities. Its core is a robust self-tuning data association that adapts the search area depending on a pseudo-entropy of the measurements. Additionally, to smooth the final result, we adjust the information matrix for the associated data as a function of the relative transform produced by the data association process. We evaluate our method on real data from urban and rural scenarios around the city of Karlsruhe in Germany. We compare state-of-the-art outlier mitigation methods with our self-tuning approach, demonstrating a considerable improvement, especially for outer-urban scenarios.
- 25 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 7 (4) , 12339-12346
Robust Self-Tuning Data Association for Geo-Referencing Using Lane Markings
Localization in aerial imagery-based maps offers many advantages, such as global consistency, geo-referenced maps, and the availability of publicly accessible data. However, the landmarks that can be observed from both aerial imagery and on-board sensors is limited. This leads to ambiguities or aliasing during the data association. Building upon a highly informative representation (that allows efficient data association), this paper presents a complete pipeline for resolving these ambiguities. Its core is a robust self-tuning data association that adapts the search area depending on a pseudo-entropy of the measurements. Additionally, to smooth the final result, we adjust the information matrix for the associated data as a function of the relative transform produced by the data association process. We evaluate our method on real data from urban and rural scenarios around the city of Karlsruhe in Germany. We compare state-of-the-art outlier mitigation methods with our self-tuning approach, demonstrating a considerable improvement, especially for outer-urban scenarios.
- 30 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 23 (12) , 24428-24438
OpenStreetMap-Based Autonomous Navigation With LiDAR Naive-Valley-Path Obstacle Avoidance
OpenStreetMaps (OSM) is currently studied as the environment representation for autonomous navigation. It provides advantages such as global consistency, a heavy-less map construction process, and a wide variety of road information publicly available. However, the location of this information is usually not very accurate locally. In this paper, we present a complete autonomous navigation pipeline using OSM information as environment representation for global planning. To avoid the flaw of local low-accuracy, we offer the novel LiDAR-based Naive-Valley-Path (NVP) method that exploits the concept of “valley” areas to infer the local path always furthest from obstacles. This behavior allows navigation always through the center of trafficable areas following the road’s shape independently of OSM error. Furthermore, NVP is a naive method that is highly sample-time-efficient. This time efficiency also enables obstacle avoidance, even for dynamic objects. We demonstrate the system’s robustness in our research platform BLUE, driving autonomously across the University of Alicante Scientific Park for more than 20 km with 0.24 meters of average error against the road’s center with a 19.8 ms of average sample time. Our vehicle avoids static obstacles in the road and even dynamic ones, such as vehicles and pedestrians.
- 30 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 23 (12) , 24428-24438
OpenStreetMap-Based Autonomous Navigation With LiDAR Naive-Valley-Path Obstacle Avoidance
OpenStreetMaps (OSM) is currently studied as the environment representation for autonomous navigation. It provides advantages such as global consistency, a heavy-less map construction process, and a wide variety of road information publicly available. However, the location of this information is usually not very accurate locally. In this paper, we present a complete autonomous navigation pipeline using OSM information as environment representation for global planning. To avoid the flaw of local low-accuracy, we offer the novel LiDAR-based Naive-Valley-Path (NVP) method that exploits the concept of “valley” areas to infer the local path always furthest from obstacles. This behavior allows navigation always through the center of trafficable areas following the road’s shape independently of OSM error. Furthermore, NVP is a naive method that is highly sample-time-efficient. This time efficiency also enables obstacle avoidance, even for dynamic objects. We demonstrate the system’s robustness in our research platform BLUE, driving autonomously across the University of Alicante Scientific Park for more than 20 km with 0.24 meters of average error against the road’s center with a 19.8 ms of average sample time. Our vehicle avoids static obstacles in the road and even dynamic ones, such as vehicles and pedestrians.
- Open Access
OpenStreetMap-based Autonomous Navigation With LiDAR Naive-Valley-Path Obstacle Avoidance
OpenStreetMaps (OSM) is currently studied as the environment representation for autonomous navigation. It provides advantages such as global consistency, a heavy-less map construction process, and a wide variety of road information publicly available. However, the location of this information is usually not very accurate locally. In this paper, we present a complete autonomous navigation pipeline using OSM information as environment representation for global planning. To avoid the flaw of local low-accuracy, we offer the novel LiDAR-based Naive-Valley-Path (NVP) method that exploits the concept of "valley" areas to infer the local path always furthest from obstacles. This behavior allows navigation always through the center of trafficable areas following the road's shape independently of OSM error. Furthermore, NVP is a naive method that is highly sample-time-efficient. This time efficiency also enables obstacle avoidance, even for dynamic objects. We demonstrate the system's robustness in our research platform BLUE, driving autonomously across the University of Alicante Scientific Park for more than 20 km with 0.24 meters of average error against the road's center with a 19.8 ms of average sample time. Our vehicle avoids static obstacles in the road and even dynamic ones, such as vehicles and pedestrians.
Open AccessOpenStreetMap-based Autonomous Navigation With LiDAR Naive-Valley-Path Obstacle Avoidance
OpenStreetMaps (OSM) is currently studied as the environment representation for autonomous navigation. It provides advantages such as global consistency, a heavy-less map construction process, and a wide variety of road information publicly available. However, the location of this information is usually not very accurate locally. In this paper, we present a complete autonomous navigation pipeline using OSM information as environment representation for global planning. To avoid the flaw of local low-accuracy, we offer the novel LiDAR-based Naive-Valley-Path (NVP) method that exploits the concept of "valley" areas to infer the local path always furthest from obstacles. This behavior allows navigation always through the center of trafficable areas following the road's shape independently of OSM error. Furthermore, NVP is a naive method that is highly sample-time-efficient. This time efficiency also enables obstacle avoidance, even for dynamic objects. We demonstrate the system's robustness in our research platform BLUE, driving autonomously across the University of Alicante Scientific Park for more than 20 km with 0.24 meters of average error against the road's center with a 19.8 ms of average sample time. Our vehicle avoids static obstacles in the road and even dynamic ones, such as vehicles and pedestrians.
- Open Access
DA-LMR: A Robust Lane Marking Representation for Data Association
While complete localization approaches are widely studied in the literature, their data association and data representation subprocesses usually go unnoticed. However, both are a key part of the final pose estimation. In this work, we present DA-LMR (Delta-Angle Lane Marking Representation), a robust data representation in the context of localization approaches. We propose a representation of lane markings that encodes how a curve changes in each point and includes this information in an additional dimension, thus providing a more detailed geometric structure description of the data. We also propose DC-SAC (Distance-Compatible Sample Consensus), a data association method. This is a heuristic version of RANSAC that dramatically reduces the hypothesis space by distance compatibility restrictions. We compare the presented methods with some state-of-the-art data representation and data association approaches in different noisy scenarios. The DA-LMR and DC-SAC produce the most promising combination among those compared, reaching 98.1% in precision and 99.7% in recall for noisy data with 0.5 m of standard deviation.
Open AccessDA-LMR: A Robust Lane Marking Representation for Data Association
While complete localization approaches are widely studied in the literature, their data association and data representation subprocesses usually go unnoticed. However, both are a key part of the final pose estimation. In this work, we present DA-LMR (Delta-Angle Lane Marking Representation), a robust data representation in the context of localization approaches. We propose a representation of lane markings that encodes how a curve changes in each point and includes this information in an additional dimension, thus providing a more detailed geometric structure description of the data. We also propose DC-SAC (Distance-Compatible Sample Consensus), a data association method. This is a heuristic version of RANSAC that dramatically reduces the hypothesis space by distance compatibility restrictions. We compare the presented methods with some state-of-the-art data representation and data association approaches in different noisy scenarios. The DA-LMR and DC-SAC produce the most promising combination among those compared, reaching 98.1% in precision and 99.7% in recall for noisy data with 0.5 m of standard deviation.
- Open Access
- 4 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 143692-143705
Targetless Camera-LiDAR Calibration in Unstructured Environments
The camera-Lidar sensor fusion plays an important role in autonomous navigation research. Nowadays, the automatic calibration of these sensors remains a significant challenge in mobile robotics. In this paper, we present a novel calibration method that achieves an accurate six-degree-of-freedom (6-DOF) rigid-body transformation estimation (aka extrinsic parameters) between the camera and LiDAR sensors. This method consists of a novel co-registration approach that uses local edge features in arbitrary environments to get 3D-to-2D errors between the data of both, camera and LiDAR. Once we have 3D-to-2D errors, we estimate the relative transform, i.e., the extrinsic parameters, that minimizes these errors. In order to find the best transform solution, we use the perspective-three-point (P3P) algorithm. To refine the final calibration, we use a Kalman Filter, which gives the system high stability against noise disturbances. The presented method does not require, in any case, an artificial target, or a structured environment, and therefore, it is a target-less calibration. Furthermore, the method we present in this paper does not require to achieve a dense point cloud, which holds the advantage of not needing a scan accumulation. To test our approach, we use the state-of-the-art Kitti dataset, taking the calibration provided by the dataset as the ground truth. In this way, we achieve accuracy results, and we demonstrate the robustness of the system against very noisy observations.
- 4 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 143692-143705
Open AccessTargetless Camera-LiDAR Calibration in Unstructured Environments
The camera-Lidar sensor fusion plays an important role in autonomous navigation research. Nowadays, the automatic calibration of these sensors remains a significant challenge in mobile robotics. In this paper, we present a novel calibration method that achieves an accurate six-degree-of-freedom (6-DOF) rigid-body transformation estimation (aka extrinsic parameters) between the camera and LiDAR sensors. This method consists of a novel co-registration approach that uses local edge features in arbitrary environments to get 3D-to-2D errors between the data of both, camera and LiDAR. Once we have 3D-to-2D errors, we estimate the relative transform, i.e., the extrinsic parameters, that minimizes these errors. In order to find the best transform solution, we use the perspective-three-point (P3P) algorithm. To refine the final calibration, we use a Kalman Filter, which gives the system high stability against noise disturbances. The presented method does not require, in any case, an artificial target, or a structured environment, and therefore, it is a target-less calibration. Furthermore, the method we present in this paper does not require to achieve a dense point cloud, which holds the advantage of not needing a scan accumulation. To test our approach, we use the state-of-the-art Kitti dataset, taking the calibration provided by the dataset as the ground truth. In this way, we achieve accuracy results, and we demonstrate the robustness of the system against very noisy observations.