Songming Chen

Dr. Songming Chen


Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 July 2024 in Measurement Science and Technology
Verified
Journal Article
Closed access
MGMeng Guan
Meng Guan
ZHZhiLong Hao
ZhiLong Hao
SZShitong Zhang
Shitong Zhang
SCSongming Chen
Songming Chen
In recent years, the increasing demand for precise and intelligent systems has been driven by advancements in indoor unmanned navigation and positioning technologies. Nonetheless, the indoor environment poses challenges as objects often intersect or obstruct each other due to spatial constraints, significantly impeding sensor recognition capabilities and compromising localization accuracy. This paper presented a multi-sensor fusion framework within a system primarily comprised of LiDAR, complemented by RGB-D and IMU sensors, to mitigate these challenges. To mitigate point cloud misalignment arising from occlusions, an improved PL-ICP algorithm have been proposed,resulting in improved accuracy and speed in occlusion scenarios. Meanwhile, this paper enhanced the Otsu algorithm by leveraging RGB-D matching to extract additional feature information, consequently enhancing the system’s convergence when occlusion happens. Ultimately, the Extended Kalman Filter (EKF) algorithm is employed to fuse point cloud and image data. Extensive experimental results demonstrate that the proposed approach not only enhances localization accuracy and stability but also exhibits superior convergence characteristics, offering an economical and efficient solution for robotic navigation.

27 March 2023 in IEEE Sensors Journal
Verified
Journal Article
Closed access
SCSongming Chen
Songming Chen
Vincent FrémontVincent Frémont
Vincent Frémont
Iterative Closest Point (ICP) algorithms are widely used in the literature for the estimation of relative transformations using 3D LiDAR point clouds. This class of algorithms proves to be efficient when the 3D data share sufficient overlapping parts and a good initial guess is provided. However, large relative motions and mutual occlusion of objects in real-road scenarios hinder traditional optimization-based ICP from achieving optimal estimation. This paper explores both direct and feature-based 3D LiDAR scan matching using the ICP framework in different contexts, such as parking, residential, urban, and highway scenarios. In order to guarantee the scan matching performances in scenarios with scarce geometric information and fast ego-vehicle motion, we propose an adaptive semi-direct scan matching method together with an alignment uncertainty quantification. The proposed semi-direct scan matching is tested on both the public KITTI and self-recorded LS2N datasets, which accomplishes the robust 6 Degrees of Freedom (DoF) pose estimation and consistent scene reconstruction. We demonstrate that the proposed approach outperforms the state-of-the-art and achieves the leading results with 68.3% average relative fitness and 5.71 cm average RMSE, respectively.

22 July 2025 in Electronics
Verified
Journal Article
Open access
PDF file
XSXiaoxiao Song
Xiaoxiao Song
SCSongming Chen
Songming Chen
QLQiang Liu
Qiang Liu
This paper proposes a constrained trajectory optimization framework for autonomous vehicles (AVs) based on convex programming techniques. An enhanced kinematic vehicle model is introduced to capture dynamic motion characteristics that are often overlooked in conventional models. For obstacle avoidance, environmental constraints are transformed into convex formulations using free-space corridor methods. The trajectory planning process is further optimized through a linearized model predictive control (MPC) scheme, which considers both vehicle dynamics and environmental safety. The resulting formulation enables efficient convex optimization suitable for real-time implementation. Experimental results in various scenarios demonstrate improvements in both trajectory smoothness and safety. Furthermore, the proposed optimization method reduces the average execution time by nearly 70% compared to the nonlinear alternative, validating its computational efficiency and practical applicability.

7 April 2026
Verified
Thesis
Closed access
SCSongming Chen
Songming Chen
Perception et estimation d’état basées sur plusieurs capteurs pour les véhicules autonomes: Percevoir ou comprendre les environnements environnants est indispensable pour construire des systèmes d’aide à la conduite ou des véhicules autonomes. Dans cette thèse, nous étudions l’approche de fusion de capteurs pour le problème de localisation et de cartographie simultanées (SLAM) avec des capteurs visuels et de distance complémentaires. Afin de prendre des décisionsconservatrices et d’augmenter la sécurité de manoeuvre des véhicules autonomes, l’analyse d’incertitude de l’estimation de la pose est également mise en oeuvre. Le système SLAM traditionnel suppose des scènes statiques, ce qui est vulnérable dans le contexte d’environnements extérieurs dynamiques. Ainsi, nous introduisons une approche basée sur les données pour exploiter les informations sémantiques qui interprètent la séquence de mesure à travers les cadres, ce qui distingue efficacement les objets en mouvement des objets statiques. Nous testons les algorithmes proposé sur des données réelles de trafic urbain et d’aires de stationnement, qui présente des résultats prometteurs.

24 July 2024 in arXiv
Verified
Preprint
Open access
PDF file
NPNan Peng
Nan Peng
XZXun Zhou
Xun Zhou
MWMingming Wang
Mingming Wang
XYXiaojun Yang
Xiaojun Yang
SCSongming Chen
Songming Chen
GCGuisong Chen
Guisong Chen
Temporal information is crucial for detecting occluded instances. Existing temporal representations have progressed from BEV or PV features to more compact query features. Compared to these aforementioned features, predictions offer the highest level of abstraction, providing explicit information. In the context of online vectorized HD map construction, this unique characteristic of predictions is potentially advantageous for long-term temporal modeling and the integration of map priors. This paper introduces PrevPredMap, a pioneering temporal modeling framework that leverages previous predictions for constructing online vectorized HD maps. We have meticulously crafted two essential modules for PrevPredMap: the previous-predictions-based query generator and the dynamic-position-query decoder. Specifically, the previous-predictions-based query generator is designed to separately encode different types of information from previous predictions, which are then effectively utilized by the dynamic-position-query decoder to generate current predictions. Furthermore, we have developed a dual-mode strategy to ensure PrevPredMap's robust performance across both single-frame and temporal modes. Extensive experiments demonstrate that PrevPredMap achieves state-of-the-art performance on the nuScenes and Argoverse2 datasets. Code will be available at https://github.com/pnnnnnnn/PrevPredMap.