The accurate perception of external environment information through the robot foot is crucial for the mobile robot to evaluate its ability to traverse terrain. Adequate foot-end contact signals can provide robust support for robot motion control and decision-making processes. The shape and uncertain rotation of the wheel-legged robot foot end represent a significant challenge to sensing the robot foot-end contact state, which current foot-end sensing schemes cannot solve. This paper presents a sensing method for the tire stress field of wheel-legged robots. A finite element analysis was conducted to study the deformation characteristics of the foot-end tire under force. Based on this analysis, a heuristic contact position estimator was designed that utilizes symmetrical deformation characteristics. Strain sensors, arranged in an array, extract the deformation information on the inner surface of the tire at a frequency of 200 Hz. The contact position estimator reduces the dimensionality of the data and fits the eigenvalues to the estimated contact position. Using support vector regression, the force estimator utilizes the estimated contact position and sensor signal to estimate the normal reaction force, designated as F_Z. The sensing system is capable of detecting the contact position on the wheel circumference (with a root mean square error of 1.150°), as well as the normal force of 160 N on the Z axis (with a root mean square error of 6.04%). To validate the efficacy of the sensor detection method, a series of randomized and repeated experiments were conducted on a self-constructed test platform. This novel approach offers a promising avenue for perceiving contact states in wheel-legged robots.

Acquiring terrain information during robot locomotion is essential for autonomous navigation, gait selection, and trajectory planning. Quadruped robots, due to their biomimetic structures, demonstrate enhanced traversability over complex terrains compared to other robotic platforms. Furthermore, the internal sensors of quadruped robots acquire rich terrain-related data during locomotion across diverse terrains. This study investigates the relationship between terrain characteristics and quadruped robots based on proprioception sensor data, and proposes a simple, efficient, and motion-independent terrain classification method by integrating multiple sensor signals. The sensors referred to in the text only include the IMU sensor and joint encoders, which means that the method has a wide range of applicability while requiring sufficiently low hardware cost. The Convolutional Neural Network will serve as the backbone of the algorithm. In addition, the control command about its own control information will serve as supporting information to eliminate the impact of motion patterns on the results. Employing a multi-label classification algorithm, the complex terrains are classified by multiple physical feature labels like roughness, slippage, softness, and slope, which depict terrain attributes. A feature-labeled terrain dataset is established by abstracting diverse terrain features across various terrains. Unlike semantic labels (e.g., grassland, sand, gravel) that are comprehensible only to humans, feature labels provide a more helpful and precise terrain characterization, including broader terrain attributes.

In unstructured environments, terrain perception is essential for stability and environmental awareness of Quadruped robot locomotion. Existing approaches primarily rely on visual or proprioceptive signals, but their effectiveness is limited under conditions of visual occlusion or ambiguous terrain features. To address this, this study proposes a multimodal terrain perception method that integrates acoustic features with proprioceptive signals. This terrain perception method collects environmental acoustic information through an externally mounted sound sensor, and combines the sound signal with proprioceptive sensor data from IMU and joint encoder of the quadruped robot. The method was deployed on the quadruped robot Lite2 platform developed by Deep Robotics, and experiments were conducted on four representative terrain types: concrete, gravel, sand, and carpet. Mel-spectrogram features are extracted from the acoustic signals and concatenated with the IMU and joint encoder to form feature vectors, which are subsequently fed into a support vector machine for terrain classification. For each terrain type, 400 s of data were collected. Experimental results show that the terrain classification accuracy reaches 78.28% without using acoustic signals, while increasing to 82.52% when acoustic features are incorporated. To further enhance the classification performance, this study performs a combined exploration of the SVM hyperparameters $C$ and $\gamma$ as well as the time-window length $w_{in}$. The final results demonstrate that the classification accuracy can be improved to as high as 99.53% across all four terrains.

Wheel-legged robots combine the advantages of both wheeled robots and traditional quadruped robots, enhancing terrain adaptability but posing higher demands on the perception of foot–ground contact forces. However, existing approaches still suffer from limited accuracy in estimating contact positions and three-dimensional contact forces when dealing with flexible tire–ground interactions. To address this challenge, this study proposes a foot–ground contact state detection technique and optimization method based on multi-sensor fusion and intelligent modeling for wheel-legged robots. First, finite element analysis (FEA) is used to simulate strain distribution under various contact conditions. Combined with global sensitivity analysis (GSA), the optimal placement of PVDF sensors is determined and experimentally validated. Subsequently, under dynamic gait conditions, data collected from the PVDF sensor array are used to predict three-dimensional contact forces through Gaussian process regression (GPR) and artificial neural network (ANN) models. A custom experimental platform is developed to replicate variable gait frequencies and collect dynamic contact data for validation. The results demonstrate that both GPR and ANN models achieve high accuracy in predicting dynamic 3D contact forces, with normalized root mean square error (NRMSE) as low as 8.04%. The models exhibit reliable repeatability and generalization to novel inputs, providing robust technical support for stable contact perception and motion decision-making in complex environments.

Robotic visual perception and terrain recognition are critical for autonomous locomotion and adaptive control in complex environments. However, existing models often extract weak features, confuse classes, and deliver unstable recognition. Most prior studies use end-to-end convolutional networks or single-stream feature extraction, which limits the balance between fine-grained visual representation and adaptive discrimination of confusing samples. To solve this problem, this paper proposes a vision model that blends attention mechanisms with a confusion augmentation strategy. Using an improved ResNet50 backbone, we add a local feature sharpening module and a channel–spatial attention module to strengthen edge texture and global context representation. We also design a confusion augmentation strategy based on the similarity of hard samples. It generates mixed samples through cross-perturbation in feature space, thereby improving the discrimination of highly similar terrains. Experiments show that our model achieves an accuracy of over 98.19% on various terrains, including cement, asphalt, sand, and snow. t-SNE visualization and Grad-CAM analysis demonstrate clear class separability and good interpretability, confirming the effectiveness and robustness of the approach for robotic terrain recognition.