Zhenzhong Jia
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
Research Interests
Zhenzhong Jia has a total of 65 co-authors. Here are the top co-authors by number of shared publications:
- Chen Yao
Shenzhen Key Laboratory of Biomimetic Robotics and Intelligent Systems, Department of Mechanical and Energy Engineering, Southern University of Science and Technology (SUSTech), Shenzhen, China
8 shared publications - Zheng Zhu
Shenzhen Key Laboratory of Biomimetic Robotics and Intelligent Systems, Department of Mechanical and Energy Engineering, Southern University of Science and Technology, Shenzhen, China
8 shared publications - Shipeng Lyu
Department of Mechanical Engineering, The Hong Kong Polytechnic University, Kowloon, Hong Kong
5 shared publications - Feng Xue
Department of Mechanical and Energy Engineering, Shenzhen Key Laboratory of Biomimetic Robotics and Intelligent Systems, Southern University of Science and Technology, Shenzhen, China
3 shared publications - Guowei Shi
Shenzhen Key Laboratory of Biomimetic Robotics and Intelligent Systems, Southern University of Science and Technology, Shenzhen, China
3 shared publications - Jing Wu
Department of Mechanical and Energy Engineering, Southern University of Science and Technology (SUSTech), Shenzhen, China
3 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Wiley in Advanced Intelligent Systems
TierTac: Physically‐Grounded Hierarchical Reasoning for Generalizable Tactile Perception
Tactile perception is essential for intelligent robotic systems interacting with the physical world, yet achieving reliable understanding across heterogeneous sensors remains challenging. Existing approaches rely on representation learning or independent attribute prediction, often neglecting the physically grounded hierarchical dependencies underlying contact interactions. Here, we propose TierTac, a physically grounded hierarchical reasoning framework for generalizable tactile perception. TierTac organizes tactile understanding using a physics‐informed dependency hierarchy comprising three levels: intrinsic physical properties, compliance responses, and contact states. It combines contact‐aware, sensor‐agnostic representation learning with hierarchical neuro‐symbolic reasoning to enable physically consistent multiattribute inference. Experiments across multiple datasets demonstrate improved cross‐sensor alignment, physically consistent prediction, and generalization to unseen sensors. Real‐world robotic evaluations further show more stable and physically consistent interactions under both classical control and learned policies. These results highlight physically grounded hierarchical reasoning as an effective principle for developing generalizable tactile intelligence in robotic systems.
- 9 November 2026
- Published by Wiley in Advanced Intelligent Systems
Open AccessTierTac: Physically‐Grounded Hierarchical Reasoning for Generalizable Tactile Perception
Tactile perception is essential for intelligent robotic systems interacting with the physical world, yet achieving reliable understanding across heterogeneous sensors remains challenging. Existing approaches rely on representation learning or independent attribute prediction, often neglecting the physically grounded hierarchical dependencies underlying contact interactions. Here, we propose TierTac, a physically grounded hierarchical reasoning framework for generalizable tactile perception. TierTac organizes tactile understanding using a physics‐informed dependency hierarchy comprising three levels: intrinsic physical properties, compliance responses, and contact states. It combines contact‐aware, sensor‐agnostic representation learning with hierarchical neuro‐symbolic reasoning to enable physically consistent multiattribute inference. Experiments across multiple datasets demonstrate improved cross‐sensor alignment, physically consistent prediction, and generalization to unseen sensors. Real‐world robotic evaluations further show more stable and physically consistent interactions under both classical control and learned policies. These results highlight physically grounded hierarchical reasoning as an effective principle for developing generalizable tactile intelligence in robotic systems.
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (9) , 10712-10719
FAST-LIEO2: Fast and Tightly-Coupled LiDAR-Inertial-Event Odometry
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (9) , 10712-10719
FAST-LIEO2: Fast and Tightly-Coupled LiDAR-Inertial-Event Odometry
- 18 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (3) , 3940-3956
Collision Dynamics Model and Self-Learning Control for 4WIS Vehicles
Vehicles at high speeds are prone to losing control due to the initial impact, which can potentially lead to more severe secondary collisions. An effective post-impact control (PIC) strategy can restore the vehicle's stability after the initial collision, reducing the risk of secondary collisions. This paper proposes a robust four-wheel independent steering (4WIS) PIC strategy that is based on deep reinforcement learning (DRL); it is capable of automatically adapting to different road adhesion conditions. To our best knowledge, this paper presents the first real-world experimental validation of a DRL-based PIC strategy. We propose an 8-DOF vehicle collision dynamics model to construct a reinforcement learning training environment, and design a reward function to guide the agent in controlling the vehicle to return to its target path and achieve a safe yaw angle. We randomize a specific parameter in the tire model to simulate different road adhesion conditions and enhance the robustness. The PIC strategy can restore the vehicle's post-impact stability under different speeds and road adhesion conditions, reduce lateral post-impact errors, and return the vehicle to a safe yaw angle, further lowering the risk of secondary collisions. The model's validity was confirmed by comparing the dynamic responses of the proposed model with those of the CarSim model. The effectiveness and generalizability of the proposed PIC strategy were further verified on the CarSim and MATLAB/Simulink joint simulation platform. And experiments were conducted using a specially developed down-scaled vehicle to validate the feasibility and real-time performance of the proposed PIC strategy.
- 18 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (3) , 3940-3956
Collision Dynamics Model and Self-Learning Control for 4WIS Vehicles
Vehicles at high speeds are prone to losing control due to the initial impact, which can potentially lead to more severe secondary collisions. An effective post-impact control (PIC) strategy can restore the vehicle's stability after the initial collision, reducing the risk of secondary collisions. This paper proposes a robust four-wheel independent steering (4WIS) PIC strategy that is based on deep reinforcement learning (DRL); it is capable of automatically adapting to different road adhesion conditions. To our best knowledge, this paper presents the first real-world experimental validation of a DRL-based PIC strategy. We propose an 8-DOF vehicle collision dynamics model to construct a reinforcement learning training environment, and design a reward function to guide the agent in controlling the vehicle to return to its target path and achieve a safe yaw angle. We randomize a specific parameter in the tire model to simulate different road adhesion conditions and enhance the robustness. The PIC strategy can restore the vehicle's post-impact stability under different speeds and road adhesion conditions, reduce lateral post-impact errors, and return the vehicle to a safe yaw angle, further lowering the risk of secondary collisions. The model's validity was confirmed by comparing the dynamic responses of the proposed model with those of the CarSim model. The effectiveness and generalizability of the proposed PIC strategy were further verified on the CarSim and MATLAB/Simulink joint simulation platform. And experiments were conducted using a specially developed down-scaled vehicle to validate the feasibility and real-time performance of the proposed PIC strategy.
- 1 December 2024
- Published by World Scientific Pub Co Pte Ltd in International Journal of Artificial Intelligence and Robotics Research
- Vol. 01 (03n04)
PAWS: Perception and Adaptation Walking Synthesizer for Vision-Language-Guided Quadruped Bionic Movements
Addressing deficiencies in motion diversity and long-horizon task handling in quadruped robot algorithms, this study developed the PAWS algorithm. PAWS combines human feedback reinforcement learning with vision-language models (VLMs) to create an end-to-end, one-shot motion strategy. By analyzing video frames of canine behaviors, PAWS extracts key motion features using VLMs and designs new motion strategies for quadruped robots. This innovative method allows robots to learn from internet videos and directly map features to joint actions, bypassing traditional natural language instructions.
- 1 December 2024
- Published by World Scientific Pub Co Pte Ltd in International Journal of Artificial Intelligence and Robotics Research
- Vol. 01 (03n04)
PAWS: Perception and Adaptation Walking Synthesizer for Vision-Language-Guided Quadruped Bionic Movements
Addressing deficiencies in motion diversity and long-horizon task handling in quadruped robot algorithms, this study developed the PAWS algorithm. PAWS combines human feedback reinforcement learning with vision-language models (VLMs) to create an end-to-end, one-shot motion strategy. By analyzing video frames of canine behaviors, PAWS extracts key motion features using VLMs and designs new motion strategies for quadruped robots. This innovative method allows robots to learn from internet videos and directly map features to joint actions, bypassing traditional natural language instructions.
- Open Access
Ubiquitous Field Transportation Robots with Robust Wheel-Leg Transformable Modules
This paper introduces two field transportation robots. Both robots are equipped with transformable wheel-leg modules, which can smoothly switch between operation modes and can work in various challenging terrains. SWhegPro, with six S-shaped legs, enables transporting loads in challenging uneven outdoor terrains. SWhegPro3, featuring four three-impeller wheels, has surprising stair-climbing performance in indoor scenarios. Different from ordinary gear-driven transformable mechanisms, the modular wheels we designed driven by self-locking electric push rods can switch modes accurately and stably with high loads, significantly improving the load capacity of the robot in leg mode. This study analyzes the robot's wheel-leg module operation when the terrain parameters change. Through the derivation of mathematical models and calculations based on simplified kinematic models, a method for optimizing the robot parameters and wheel-leg structure parameters is finally proposed.The design and control strategy are then verified through simulations and field experiments in various complex terrains, and the working performance of the two field transportation robots is calculated and analyzed by recording sensor data and proposing evaluation methods.
Open AccessUbiquitous Field Transportation Robots with Robust Wheel-Leg Transformable Modules
This paper introduces two field transportation robots. Both robots are equipped with transformable wheel-leg modules, which can smoothly switch between operation modes and can work in various challenging terrains. SWhegPro, with six S-shaped legs, enables transporting loads in challenging uneven outdoor terrains. SWhegPro3, featuring four three-impeller wheels, has surprising stair-climbing performance in indoor scenarios. Different from ordinary gear-driven transformable mechanisms, the modular wheels we designed driven by self-locking electric push rods can switch modes accurately and stably with high loads, significantly improving the load capacity of the robot in leg mode. This study analyzes the robot's wheel-leg module operation when the terrain parameters change. Through the derivation of mathematical models and calculations based on simplified kinematic models, a method for optimizing the robot parameters and wheel-leg structure parameters is finally proposed.The design and control strategy are then verified through simulations and field experiments in various complex terrains, and the working performance of the two field transportation robots is calculated and analyzed by recording sensor data and proposing evaluation methods.
- 31 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (7) , 6720-6727
Foot Vision: A Vision-Based Multi-Functional Sensorized Foot for Quadruped Robots
Quadruped robots equipped with sensorless feet can only provide limited information regarding the foot interaction with the surroundings, limiting their off-road exploration capability in unstructured environments. To tackle this problem, we present Foot Vision, an innovative vision-based sensorized foot for quadruped robots. It consists of a compliant ankle, a transparent foot sole with visual markers, and an internal camera. This foot is designed to be low-cost, lightweight, and adaptable to both deformable and rigid terrains. The camera can efficiently track marker displacements to estimate the foot pose. We extract the foot deformation features from the foot pose and develop a learning-based method to reconstruct 6D contact forces leveraging these features. Representative loading tests validate the foot's ability to accurately $(R^{2}>0.95)$ and dynamically predict contact forces and torques in real-time $(\text{100}~Hz)$ . Installed on the quadruped robot, the sensorized foot demonstrates its capabilities to perceive 6D contact forces on terrains with varying stiffness and estimate the local inclination of the contact surface. Furthermore, thanks to the transparent foot sole, the in-foot camera can directly identify the terrain types and measure the soil flow distribution. Experimental results confirm that the proposed vision-based sensorized foot can enhance the contact perception capability of legged robots in diverse environments.
- 31 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (7) , 6720-6727
Foot Vision: A Vision-Based Multi-Functional Sensorized Foot for Quadruped Robots
Quadruped robots equipped with sensorless feet can only provide limited information regarding the foot interaction with the surroundings, limiting their off-road exploration capability in unstructured environments. To tackle this problem, we present Foot Vision, an innovative vision-based sensorized foot for quadruped robots. It consists of a compliant ankle, a transparent foot sole with visual markers, and an internal camera. This foot is designed to be low-cost, lightweight, and adaptable to both deformable and rigid terrains. The camera can efficiently track marker displacements to estimate the foot pose. We extract the foot deformation features from the foot pose and develop a learning-based method to reconstruct 6D contact forces leveraging these features. Representative loading tests validate the foot's ability to accurately $(R^{2}>0.95)$ and dynamically predict contact forces and torques in real-time $(\text{100}~Hz)$ . Installed on the quadruped robot, the sensorized foot demonstrates its capabilities to perceive 6D contact forces on terrains with varying stiffness and estimate the local inclination of the contact surface. Furthermore, thanks to the transparent foot sole, the in-foot camera can directly identify the terrain types and measure the soil flow distribution. Experimental results confirm that the proposed vision-based sensorized foot can enhance the contact perception capability of legged robots in diverse environments.
- 30 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (7) , 6696-6703
TAIL: A Terrain-Aware Multi-Modal SLAM Dataset for Robot Locomotion in Deformable Granular Environments
Terrain-aware perception holds the potential to improve the robustness and accuracy of autonomous robot navigation in the wilds, thereby facilitating effective off-road traversals. However, the lack of multi-modal perception across various motion patterns hinders the solutions of Simultaneous Localization And Mapping (SLAM), especially when confronting non-geometric hazards in demanding landscapes. In this paper, we first propose a Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains. It incorporates various types of robotic proprioception and distinct ground interactions for the unique challenges and benchmark of multi-sensor fusion SLAM. The versatile sensor suite comprises stereo frame cameras, multiple ground-pointing RGB-D cameras, a rotating 3D LiDAR, an IMU, and an RTK device. This ensemble is hardware-synchronized, well-calibrated, and self-contained. Utilizing both wheeled and quadrupedal locomotion, we efficiently collect comprehensive sequences to capture rich unstructured scenarios. It spans the spectrum of scope, terrain interactions, scene changes, ground-level properties, and dynamic robot characteristics. We benchmark several state-of-the-art SLAM methods against ground truth and provide performance validations. Corresponding challenges and limitations are also reported.
- 30 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (7) , 6696-6703
TAIL: A Terrain-Aware Multi-Modal SLAM Dataset for Robot Locomotion in Deformable Granular Environments
Terrain-aware perception holds the potential to improve the robustness and accuracy of autonomous robot navigation in the wilds, thereby facilitating effective off-road traversals. However, the lack of multi-modal perception across various motion patterns hinders the solutions of Simultaneous Localization And Mapping (SLAM), especially when confronting non-geometric hazards in demanding landscapes. In this paper, we first propose a Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains. It incorporates various types of robotic proprioception and distinct ground interactions for the unique challenges and benchmark of multi-sensor fusion SLAM. The versatile sensor suite comprises stereo frame cameras, multiple ground-pointing RGB-D cameras, a rotating 3D LiDAR, an IMU, and an RTK device. This ensemble is hardware-synchronized, well-calibrated, and self-contained. Utilizing both wheeled and quadrupedal locomotion, we efficiently collect comprehensive sequences to capture rich unstructured scenarios. It spans the spectrum of scope, terrain interactions, scene changes, ground-level properties, and dynamic robot characteristics. We benchmark several state-of-the-art SLAM methods against ground truth and provide performance validations. Corresponding challenges and limitations are also reported.
- Open Access
TAIL: A Terrain-Aware Multi-Modal SLAM Dataset for Robot Locomotion in Deformable Granular Environments
Terrain-aware perception holds the potential to improve the robustness and accuracy of autonomous robot navigation in the wilds, thereby facilitating effective off-road traversals. However, the lack of multi-modal perception across various motion patterns hinders the solutions of Simultaneous Localization And Mapping (SLAM), especially when confronting non-geometric hazards in demanding landscapes. In this paper, we first propose a Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains. It incorporates various types of robotic proprioception and distinct ground interactions for the unique challenges and benchmark of multi-sensor fusion SLAM. The versatile sensor suite comprises stereo frame cameras, multiple ground-pointing RGB-D cameras, a rotating 3D LiDAR, an IMU, and an RTK device. This ensemble is hardware-synchronized, well-calibrated, and self-contained. Utilizing both wheeled and quadrupedal locomotion, we efficiently collect comprehensive sequences to capture rich unstructured scenarios. It spans the spectrum of scope, terrain interactions, scene changes, ground-level properties, and dynamic robot characteristics. We benchmark several state-of-the-art SLAM methods against ground truth and provide performance validations. Corresponding challenges and limitations are also reported. All associated resources are accessible upon request at \url{https://tailrobot.github.io/}.
Open AccessTAIL: A Terrain-Aware Multi-Modal SLAM Dataset for Robot Locomotion in Deformable Granular Environments
Terrain-aware perception holds the potential to improve the robustness and accuracy of autonomous robot navigation in the wilds, thereby facilitating effective off-road traversals. However, the lack of multi-modal perception across various motion patterns hinders the solutions of Simultaneous Localization And Mapping (SLAM), especially when confronting non-geometric hazards in demanding landscapes. In this paper, we first propose a Terrain-Aware multI-modaL (TAIL) dataset tailored to deformable and sandy terrains. It incorporates various types of robotic proprioception and distinct ground interactions for the unique challenges and benchmark of multi-sensor fusion SLAM. The versatile sensor suite comprises stereo frame cameras, multiple ground-pointing RGB-D cameras, a rotating 3D LiDAR, an IMU, and an RTK device. This ensemble is hardware-synchronized, well-calibrated, and self-contained. Utilizing both wheeled and quadrupedal locomotion, we efficiently collect comprehensive sequences to capture rich unstructured scenarios. It spans the spectrum of scope, terrain interactions, scene changes, ground-level properties, and dynamic robot characteristics. We benchmark several state-of-the-art SLAM methods against ground truth and provide performance validations. Corresponding challenges and limitations are also reported. All associated resources are accessible upon request at \url{https://tailrobot.github.io/}.
- 26 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 29 (6) , 4039-4050
STAF: Interaction-Based Design and Evaluation of Sensorized Terrain-Adaptive Foot for Legged Robot Traversing on Soft Slopes
Legged robots have been widely used in unstructured field environments, where unforeseen nongeometric hazards, such as sinkage or slippage might compromise their mobility. Such uncertain terrain-related risks may be very challenging to mitigate by employing point or ball-shaped feet, however, determining the appropriate principles for foot design remains an open-research problem. This study presents a systematic pipeline of design and evaluations for the sensorized terrain-adaptive foot (STAF) aiming to remodel the underlying soft, sloped granular media. The advanced foot configuration enables explicit characterization of complex foot-terrain interaction responses, thereby facilitating evidence-based evaluations using proposed terrain physical metrics and terrain traveling metrics. These terrain-aware metrics are interpretable, physics-informed, and versatile for various robot prototypes. Compared to traditional empirical foot design methods, our model-based guideline can reliably identify physical terrain properties and explain contact behavior to optimize the foot sole's primary parameters. We conduct multiple experiments on both the single-foot testbed and a quadruped robot. The findings demonstrate that our developed STAF system significantly enhances mobility and in situ sensing capabilities over loose deformable slopes.
- 26 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 29 (6) , 4039-4050
STAF: Interaction-Based Design and Evaluation of Sensorized Terrain-Adaptive Foot for Legged Robot Traversing on Soft Slopes
Legged robots have been widely used in unstructured field environments, where unforeseen nongeometric hazards, such as sinkage or slippage might compromise their mobility. Such uncertain terrain-related risks may be very challenging to mitigate by employing point or ball-shaped feet, however, determining the appropriate principles for foot design remains an open-research problem. This study presents a systematic pipeline of design and evaluations for the sensorized terrain-adaptive foot (STAF) aiming to remodel the underlying soft, sloped granular media. The advanced foot configuration enables explicit characterization of complex foot-terrain interaction responses, thereby facilitating evidence-based evaluations using proposed terrain physical metrics and terrain traveling metrics. These terrain-aware metrics are interpretable, physics-informed, and versatile for various robot prototypes. Compared to traditional empirical foot design methods, our model-based guideline can reliably identify physical terrain properties and explain contact behavior to optimize the foot sole's primary parameters. We conduct multiple experiments on both the single-foot testbed and a quadruped robot. The findings demonstrate that our developed STAF system significantly enhances mobility and in situ sensing capabilities over loose deformable slopes.
- 24 January 2024
- Published by Cambridge University Press (CUP) in Robotica
- Vol. 42 (4) , 1-17
A contact parameter estimation method for multi-modal robot locomotion on deformable granular terrains
In this paper, we consider the problem of contact parameters (slippage and sinkage) estimation for multi-modal robot locomotion on granular terrains. To describe the contact events in the same framework for robots operated at different modes (e.g., wheel, leg), we propose a unified description of contact parameters for multi-modal robots. We also provide a parameter estimation method for multi-modal robots based on CNN and DWT (discrete wavelet transformation) techniques and verify its effectiveness over different types of granular terrains. Besides motion modes, this paper also considers the influence of slope angles and the robot’s handing angles over contact parameters. Through comparison and analysis of the prediction results, our method can not only effectively predict the contact parameters of multi-modal robot locomotion on a granular medium (better than accuracy) but also achieves the same or better performance when compared to other (direct) contact measurement methods designed for individual motion modes, that is, single-modal robots such as quadruped robots and mars rovers. Our proposed unified contact parameter estimation method can be useful for studying the interaction mechanics between multi-modal robots and granular terrains as well as terrain classification tasks due to its superior sensitivity which is analyzed in the experiments.
- 24 January 2024
- Published by Cambridge University Press (CUP) in Robotica
- Vol. 42 (4) , 1-17
A contact parameter estimation method for multi-modal robot locomotion on deformable granular terrains
In this paper, we consider the problem of contact parameters (slippage and sinkage) estimation for multi-modal robot locomotion on granular terrains. To describe the contact events in the same framework for robots operated at different modes (e.g., wheel, leg), we propose a unified description of contact parameters for multi-modal robots. We also provide a parameter estimation method for multi-modal robots based on CNN and DWT (discrete wavelet transformation) techniques and verify its effectiveness over different types of granular terrains. Besides motion modes, this paper also considers the influence of slope angles and the robot’s handing angles over contact parameters. Through comparison and analysis of the prediction results, our method can not only effectively predict the contact parameters of multi-modal robot locomotion on a granular medium (better than accuracy) but also achieves the same or better performance when compared to other (direct) contact measurement methods designed for individual motion modes, that is, single-modal robots such as quadruped robots and mars rovers. Our proposed unified contact parameter estimation method can be useful for studying the interaction mechanics between multi-modal robots and granular terrains as well as terrain classification tasks due to its superior sensitivity which is analyzed in the experiments.