Terrain Types Research Articles

Improving Out-of-Distribution Generalization in SAR Image Scene Classification with Limited Training Samples

For practical maritime SAR image classification tasks with special imaging platforms, scenes to be classified are often different from those in the training sets. The quantity and diversity of the available training data can also be extremely limited. This problem of out-of-distribution (OOD) generalization with limited training samples leads to a sharp drop in the performance of conventional deep learning algorithms. In this paper, a knowledge-guided neural network (KGNN) model is proposed to overcome these challenges. By analyzing the saliency features of various maritime SAR scenes, universal knowledge in descriptive sentences is summarized. A feature integration strategy is designed to assign the descriptive knowledge to the ResNet-18 backbone. Both the individual semantic information and the inherent relations of the entities in SAR images are addressed. The experimental results show that our KGNN method outperforms conventional deep learning models in OOD scenarios with varying training sample sizes and achieves higher robustness in handling distributional shifts caused by weather conditions, terrain type, and sensor characteristics. In addition, the KGNN model converges within many fewer epochs during training. The performance improvement indicates that the KGNN model learns representations guided by beneficial properties for ODD generalization with limited training samples.

EAT: Environment Agnostic Traversability for reactive navigation

This work presents EAT (Environment Agnostic Traversability for Reactive Navigation) a novel framework for traversability estimation in indoor, outdoor, subterranean (SubT) and other unstructured environments. The architecture provides updates on traversable regions online during the mission, adapts to varying environments, while being robust to noisy semantic image segmentation. The proposed framework considers terrain prioritization based on a novel decay exponential function to fuse the semantic information and geometric features extracted from RGB-D images to obtain the traversability of the scene. Moreover, EAT introduces an obstacle inflation mechanism on the traversability image, based on mean-window weighting module, allowing to adapt the proximity to untraversable regions. The overall architecture uses two LRASPP MobileNet V3 large Convolutional Neural Networks (CNN) for semantic segmentation over RGB images, where the first one classifies the terrain types and the second one classifies see-through obstacles in the scene. Additionally, the geometric features profile the underlying surface properties of the local scene, extracting normals from depth images. The proposed scheme was integrated with a control architecture in reactive navigation scenarios and was experimentally validated in indoor and outdoor environments as well as in subterranean environments with a Pioneer 3AT mobile robot.

Vibration-Based Recognition of Wheel-Terrain Interaction for Terramechanics Model Selection and Terrain Parameter Identification for Lugged-Wheel Planetary Rovers.

Identifying terrain parameters is important for high-fidelity simulation and high-performance control of planetary rovers. The wheel-terrain interaction classes (WTICs) are usually different for rovers traversing various types of terrain. Every terramechanics model corresponds to its wheel-terrain interaction class (WTIC). Therefore, for terrain parameter identification of the terramechanics model when rovers traverse various terrains, terramechanics model switching corresponding to the WTIC needs to be solved. This paper proposes a speed-independent vibration-based method for WTIC recognition to switch the terramechanics model and then identify its terrain parameters. In order to switch terramechanics models, wheel-terrain interactions are divided into three classes. Three vibration models of wheels under three WTICs have been built and analyzed. Vibration features in the models are extracted and non-dimensionalized to be independent of wheel speed. A vibration-feature-based recognition method of the WTIC is proposed. Then, the terrain parameters of the terramechanics model corresponding to the recognized WTIC are identified. Experiment results obtained using a Planetary Rover Prototype show that the identification method of terrain parameters is effective for rovers traversing various terrains. The relative errors of estimated wheel-terrain interaction force with identified terrain parameters are less than 16%, 12%, and 9% for rovers traversing hard, gravel, and sandy terrain, respectively.

Slip considered path planning of a novel angled-spoke based robot in a terrain mixture of granular media and rigid support

Navigation in terrains mixture of rigid support and granular media is associated with slippage. It is important to find the optimal path because slippage implies the possibility of a mobile robot being stuck in the sand. This research introduce concept of effective distance which consider effect of slippage to present a path planning algorithm. Effective distance is a conceptual distance that the mobile robot traverse to reach next node, and it is calculated based on traversable difficulty. Traversable difficulty varies with the slope angle, and the revolution speed of the angled spoke-based wheel (ASW) is estimated using meta-model. The meta-model is achieved empirically. Experiments have done in testbench which can implement slope angle and terrain types. Effectiveness of this algorithm was verified through simulation. For the simulations three cost determination methods, namely, the effective distance with varying revolution speed of the ASW, the effective distance with fixed revolution speed of the ASW, and actual distance were examined in four different scenarios. The effective distance with varying revolution speed of the ASW has maximum 135.94%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$135.94\\%$$\\end{document} longer total actual distance. However, it results minimum 44.72%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$44.72\\%$$\\end{document} shorter total effective distance.

Cluster analysis of cloud-to-ground lightning activities near transmission lines in typical terrains

Cluster analysis of cloud-to-ground lightning activities near transmission lines in typical terrains

Comparison and analysis of ground seed detectors and interpolation methods in airborne LiDAR filtering

Ground seed detectors and interpolation methods are fundamental components of filtering algorithms. However, the performance of different detectors and interpolation methods typically varies, and few studies have been conducted on the adaptability of different detectors and interpolation methods to different terrains. Therefore, we compare three ground seed detectors (cylindrical neighborhood (CN), fixed grid (FG), and moving window (MW)) and three interpolation methods (triangulated irregular network (TIN), thin plate spline (TPS), and inverse distance weighting (IDW)). In addition, nine filters are constructed by combining the three ground seed detectors and the three interpolation methods to evaluate their comprehensive influences. To assess the performance of these detectors, interpolation methods, and filters, fifteen ISPRS-supplied light detection and ranging (LiDAR) benchmark datasets are utilized in the experiment. The findings indicate that the CN-TPS filter, which combines the CN detector and TPS method, achieves superior performance across mean Pg (99.16 % – correctly classified ground points divided by all extracted ground points), RMSE (0.44 m – root mean squared error), and total error (3.78 %). Moreover, the filtering methods are mainly affected by the performance of the ground seed detector and are less affected by the selected interpolation method. These results can be used to provide a valuable reference for designing an optimal LiDAR filtering algorithm for varied terrain types and applications.

Estimating omnidirectional urban vertical wind speed with direction-dependent building morphologies

A main external factor that limits the energy assessment of high-rise buildings in urban areas is the vertical wind conditions. Conventional buildings' energy assessment tools apply a universal vertical wind profile estimation equation in all directions and only consider four typical terrain types. However, the various shape and configurations of buildings in cities can induce the unique wind speed in each direction, and the vertical wind speeds in the same terrain type can be different even in the same city. To get more realistic vertical wind conditions, this study developed an omnidirectional urban vertical wind speed estimation method with direction-dependent building morphologies. A pie-shaped segment method is applied to calculate building morphological features in approaching wind direction. Machine learning methods are developed to calculate parameters of the modified direction-dependent power law urban wind function to convert the Typical Meteorological Year (TMY) wind speed to urban vertical wind speeds. To validate the proposed method, wind tunnel data in Hong Kong are collected for a case study. The results indicate that the proposed method with morphological inputs can effectively estimate the vertical wind speed at different wind directions and heights, and deep neural networks and support vector regression have better performance.

Yield and growth rates of pastures grown in the Bay of Plenty region over 35 years

Bay of Plenty district, and particularly the Rotorua Lakes area, has a diverse terrain and soil types which influence pasture production. Regular measurement of pasture production enables short term decision making on a farm scale and is invaluable for catchment,district and regional long term management strategies. Thomas (Tom) M. Gee, was a retired farmer with more than 18 years of field trial experience with MAF Field Research Division. He collected data from more than 30 farms after he retired. Data from other sites in the district were collected in the early 1970s by MAF technicians stationed in Whakatane and Tauranga and later by AgResearch staff and a farm consultant based in Rotorua. Tom Gee’s mission was to use these measurements to provide farmers with rates of growth (ROG) data to inform them about their farm. The Gee farm (Fairbank) of 200 ha was originally leased from Ngati Whakaue Tribal Lands in 1916 and then purchased before much of it was sold back to the Incorporation in 1970. Tom retired in 1989 but kept meticulously recording pasture growth rates on different farms up to ~ 2007. Some field notes were lost, but datasets with gaps are still useful to assist monthly growth rates calculations. His valuable and extensive (almost 25 years) on farm field records have been retrieved, compiled, assembled, and digitised, to be saved electronically, and entered into the AgYields National Database hosted at Lincoln University. Part of this legacy dataset has been summarised and dry matter yields and growth rates calculated, consistent withprevious methods, to provide a quantified description of mean monthly pasture growth rates across the Bay of Plenty region, in New Zealand.

Precision Location-Aware and Intelligent Scheduling System for Monorail Transporters in Mountain Orchards

This study addressed the issue of the real-time monitoring and control of the transporter in a mountain orchard terrain characterized by varying topography, closed canopy, shade, and other environmental factors. This study involved independent research and the development of a series of electric monorail transporters. First, the application requirements of “Where is the monorail transporter?” were examined, and an accurate location-aware method based on high-frequency radio frequency identification (RFID) technology was proposed. In addition, a location-aware hardware system based on STM32 + RFID + LoRa was designed to determine the position of the monorail transporter on a rail. Second, regarding the application requirements of “Where is the monorail transporter going?”, a multimode control gateway system based on Raspberry Pi + LoRa + 5G was designed. An Android mobile terminal can obtain operational information about the transport plane in real time through the gateway system and remotely control its operation. The track-changing branch structure enables multimachine autonomous intelligent avoidance. Based on the experimental results of monorail transporter positioning in mountain orchards under various typical terrains, such as flat surfaces, turning paths, and uphill/downhill slopes, the road section average relative error of the 7ZDGS–250-type monorail transporter was 1.27% when the distance between benchmark positioning tags was set at 10 m on both flat and turning roads, and that of the 7ZDGS–300-type monorail transporter was 1.35% when the distance between benchmark positioning tags was set at 6 m uphill/downhill. The road section relative error of the 7ZDGS–250-type monorail transporter was 21.18%, and that of the 7ZDGS–300-type monorail transporter was 9.96%. In addition, the experimental results of monorail transporter communication control showed that the combination of the multimode control gateway control system and track-changing branch structure can achieve multimachine cooperation and autonomous avoidance function, ensuring that multiple monorail transporters can operate simultaneously without collision. The findings of this study establish the communication link of “monorail transporter-gateway system-control terminal” and form a precise positioning and real-time control scheme applicable to the operating environment of monorail transporters, thereby improving the intelligence and safety of mountain orchard monorail transporters.

Using the Geoelectric Method to Identify Groundwater Schlumberger Configuration Resistivity in the Gambesi and Sasa Villages, South Ternate District, North Maluku Province

One of the geophysical techniques that may be used to define subsurface lithology is the geoelectric approach. The Schlumberger configuration's resistivity technique is used on the assumption that the distance between the potential electrodes is fixed. In contrast, the distance between the current electrodes progressively varies following the resistivity-to-depth curve and the resistivity table. This study aims to determine the lithology of the soil layer and the lithology of the soil layer that serves as an aquifer based on their respective resistivity values. The acquired data demonstrate the presence of multiple layers, including alluvium, sand, gravel, and andesite, in the subsurface lithology at two sites with a depth of 25 meters and a track length of 50 and 100 meters. Based on the resistivity curve to depth and the findings of data processing at the two data collecting sites, groundwater at location 1 (Gambesi Village) could not be identified in the resistivity table because of the location's hilly terrain and alluvial rock types. However, at location 2 (Sasa Village), groundwater was present at depths of 12.8 meters with thicknesses of 6.24 cm and 0.57 Ωm and at depths of 12.8 meters with thicknesses of 4.91 meters and 297 Ωm, respectively. This region slopes down from this position.

Robust and adaptive terrain classification and gait event detection system

Real-time gait event detection (GED) system can be utilized for gait analysis and tracking fitness activities. GED for various types of terrains (e.g., stair-walk, uneven surfaces, etc.) is still an open research problem. This study presents an inertial sensor-based approach for real-time GED system that works for diverse terrains in an uncontrolled environment. The GED system classifies three types of terrains, i.e., flat-walk, stair-ascend and stair-descend, with an average classification accuracy of 99%. It also accurately detects various gait events, including, toe-strike, heel-rise, toe-off, and heel-strike. It is computationally efficient, implemented on a low-cost microcontroller, works in real-time and can be used in portable rehabilitation devices for use in dynamic environments.

TTA-GEP: Terrain Traversability Analysis with Geometry and Environmental Perception for the Path Planning of Planetary Rovers

Terrain traversability analysis (TTA), the key to the navigation of planetary rovers, is significant to the safety of the rover. Therefore, owing to its complexity, the Martian terrain is worth analysing comprehensively based on the terrain variability and hazard level. In this work, we propose a novel method for terrain traversability analysis for the path planning of planetary rovers by integrating Martian terrain geometry features with terrain semantic information, which includes geometry and environmental perception (GEP). Specifically, we deploy semantic segmentation to classify common terrain types, such as rocks, bedrocks, and sand, obtaining semantic information as one part of terrain traversability analysis at the same time. Simultaneously, the point cloud is generated by using binocular images from the planetary rover navigation camera (Navcam) to construct a 2.5D elevation map of the environment to analyse the geometric characteristics of the terrain. Besides, we implement path planning based on the results of TTA-GEP. Overall, our proposed method improves the performance of the terrain traversability analysis and reduces the risk of planetary rovers while detecting in an unstructured environment.

Comparison and Assessment of Different Land Cover Datasets on the Cropland in Northeast China

The provision of precise and dependable information regarding the extent and distribution of cropland is imperative for the evaluation of food security, agricultural planning, and resource management. Cropland is an important component of land cover type and is offered in multiple existing global/regional land cover products. However, global-scale accuracy evaluation may not be representative of class-specific or local-area accuracy, such as in Northeast China, which is an important grain-producing region of China and has various types of cultivated land (e.g., wheat, rice) and diverse terrains. It poses a great challenge in generating precise cropland classification by automated mapping. Thus, it is indispensable to evaluate the accuracy and reliability of these various land cover datasets before using them. In this study, we collected thirteen sets of global or national-scale land cover datasets. Through the visual interpretation of high-resolution images, ground “truth” samples were collected to evaluate the data accuracy across Northeast China. The overall accuracy (OA) evaluation results in Phase-2020 show that CLCD has the highest value with 0.914, followed by GlobeLand30 (0.906), GLC_FCS30 (0.902), and Esri (0.896) for cropland classification in Northeast China. CGLS-LC100 has the lowest OA (0.710). For the commission and omission errors of six datasets in Phase-2020, CGLS-LC100 has an obvious overestimation (larger commission error), while the two national-scale datasets (CLCD and CLUDs) perform relatively better. In terms of spatial consistency, high spatial agreement among the nine Phase-2015 datasets or in the six Phase-2020 datasets could be discovered in traditional agricultural regions like the Sanjiang–Songnen–Liaohe Plain, and low agreement is found in the transition areas of mountains (hills) and plains with the mixed landscape of forest (grassland) and farmland. In the aspect of comparison pairwise data, CLCD is in good agreement with GLC_FCS30, GlobeLand30, and Esri, while CGLS-LC100 is in the poorest agreement with any other dataset. The comparison and evaluation results are expected to provide a reference on which aspects and to what extent these land cover products may be consistent and guide the cropland data product selection for Northeast China.

Examining the Efficiency of Electric-Assisted Mountain Biking across Different Types of Terrain

Mountain bikes with electric assistance (e-bikes) have gained popularity recently by allowing riders to increase their pedaling power through an electric motor. This innovation has raised questions about how e-bikes compare to traditional mountain bikes regarding physical effort, speed, and physiological demands. By examining these factors, the study aims to compare and characterize differences in performance-related parameters when using an electric-assisted mountain bike compared to a conventional mountain bike on different types of terrain (uphill, downhill, flat section, technically demanding terrain) concerning power output, velocity, cardiorespiratory parameters, and energy expenditure. Six experienced mountain bikers (mean age: 44.6 ± 6.4 years, mean body height: 173.3 ± 5.6 cm, mean body weight: 70.6 ± 4.9 kg) cycled 4.5 km on varying off-road terrain at their own race pace, once with and once without electrical assistance, in randomized order. The results of the study indicate significantly faster (24.3 ± 1.85 to 17.2 ± 1.22 km/h (p < 0.001)) cycling on an electric-assisted mountain bike, which reduces cardiorespiratory parameters and metabolic effort as well as results in less demanding workload (138.5 ± 31.8 W) during the cycling with an electric-assisted mountain bike in comparison to a conventional mountain bike (217.5 ± 24.3 W (p < 0.001)). The results indicate significant differences especially when riding uphill. The performance advantage of an electrically assisted mountain bike diminishes compared to a conventional mountain bike on downhill, flat, or technically challenging terrain. The highlighted advantages of electric-assisted mountain bikes could represent a novel strategy for cycling in different terrains to optimize efficiency.

Enhanced flood water depth estimation from Sentinel-1A images

ABSTRACT Floods, as a natural disaster, have severe consequences for all aspects of life and society, including but not limited to damaging harvests and transportation systems, and causing epidemic diseases. Estimating flood water depths has been a topic of great interest for several decades. Finding an effective solution to this problem is crucial for accurately estimating the impact of floods and making informed decisions to mitigate their consequences. The paper presents a novel approach to estimate flood water depths in Quang Binh province, Vietnam, by utilizing an improved Otsu algorithm to classify each image pixel as either water or non-water, and then applying the FwDET interpolation algorithm to compute the flood depth of each classified water pixel. The improved algorithm can improve the accuracy of water classification when compared to the traditional Otsu method by automatically dividing images into subregions of varying sizes, which are suitable for optimal Otsu computation.The estimated results for the study area in Quang Binh province were validated using ground measurement station data and demonstrated the relatively high accuracy of water classification (ranging from 91.4% to 92%) and estimated flood water depth (85.26%). Based on the positive results obtained from the investigation of the complex mountainous area mentioned above, it can be inferred that the proposed method has the potential to be applied to estimate flood depths in various types of terrains.

Rendering Perceived Terrain Stiffness in VR Via Preload Variation Against Body-Weight.

PreloadStep is a novel platform that creates the illusion of walking on different types of terrain in Virtual Reality without requiring users to wear any special instrumentation. PreloadStep works by compressing a set of springs between two plates, with the amount of compression determining the perceived stiffness of the virtual terrain. The platform can render perception of stiffness by applying preload forces up to 824 N in different portions of the terrain, capable of changing stiffness illusion even while a user is standing on it. The effectiveness of PreloadStep was tested in two perception studies (perception thresholds and haptic-visual congruence studies) and an example application, with the results indicating that it is a promising method for creating engaging virtual terrain experiences.

Gait planning and optimization of an 18 DOF quadruped robot with compliant shanks

Legged robots made of rigid links have disadvantages like poor energy efficiency and large impact forces during foot/terrain contact while walking on 3D uneven terrain. This paper proposes a unique design of an 18 DOF quadruped robot with compliant shanks. Compliance has been added to the quadruped by introducing a “c-section“ in the shank of each leg. The robot dynamics has been modeled using the projected Newton-Euler method, while the “Craig-Bampton Method” has been utilized to model the compliant shanks. Optimal body trajectory and foot placements while maintaining dynamic stability is obtained using an NLP based optimization strategy. A torque-based inverse dynamics control technique ensures that the quadruped follows the desired trajectory. The robot was simulated using Simscape Multibody, which utilizes the robot model’s “.stl,”“.step,” and “.xml” files to follow the optimized joint trajectories. The total energy consumed and joint torques are compared between the complaint and rigid link quadruped robots for walk on different types of terrain with different gaits and C-sections of 3, 4, 6, and 8 mm thickness. The simulation results show a significant reduction in joint torque spikes (more than 70% for all cases) due to the addition of compliance in each leg. This also leads to a reduction in total energy consumption during the walk especially over an uneven terrain. Hence, the results verify that the proposed design of quadruped is functionally more efficient than its rigid counterpart for walking on uneven terrain.

The Archimede Rover: A Comparison Between Simulations and Experiments

In this paper, we propose an in-depth evaluation of the performance of the Archimede rover while traversing rough terrain with loose soil. In order to better analyze this, the reality gap is evaluated when simulating the behavior with an open-source simulator. To this extent, we implement a full model of the rover in the open-source dynamics simulator Gazebo, along with several types of terrains that replicate the experimental conditions. The rover control system is equipped with a kinematics model that allows for driving in different modes. We implement an odometric system aboard the rover, as well as an external optical absolute tracking system as reference. We estimate the drift occurring during driving in different configurations, two types of soil with corresponding wheel geometries. The results show good adherence of the odometry when the rover drives on planar ground; conversely, as expected, a marked influence of slope is seen on wheel drift. The reality gap between simulations and experimental results is kept comparatively small provided that slopes are not present.

Terrain Classification for a Quadrupedal Robot Using Proprioceptive Sensing

Quadrupedal robots can traverse a wider range of terrain types than their wheeled counterparts, but these robots do not perform the same on all terrain types. These robots are prone to undesirable behaviours like sinking and slipping on challenging terrain. To combat this issue, we can implement a terrain classifier and use the output to compute the best path for the robot to navigate. The work presented here is a terrain classifier developed for a Boston Dynamics Spot robot. The quadruped provides us with over 104 measured signals describing the position and speed of the robot and each of its four legs. The developed terrain classifier combines dimensionality reduction techniques to extract the relevant information from the signals and then applies a classification technique to differentiate terrain based on traversability. The resulting terrain classifier can identify three different terrain types with an accuracy of 93%. Ongoing work aims to get the classifier running in real time and generate cost maps to inform future path planning.

Mechanism Design and Workspace Analysis of a Hexapod Robot With Changeable Morphology

Abstract Here, the morphology of a legged robot refers to the standing/crawling pose of the robot and the forward/backward orientation of the knee, which are related to the configurations of all leg mechanisms. The standing/crawling pose of the robot determines the body height and the leg supporting region size, which are related to the obstacle-traversing and arch-traversing capability. The proper knee orientation can reduce the leg–ground interference risk and help select a good foothold. Therefore, the terrain adaptability can be enhanced if the morphology of the legged robot is changeable. Motivated by this, a hexapod robot capable of morphology conversion is designed in this study. For the leg mechanism, a double-parallelogram transmission mechanism is used to change the forward/backward orientation of the knee. The hexapod robot can transform between the crawling pose and the standing pose and can also transform among four standing morphologies (i.e., the knee–elbow, knee–knee, elbow–knee, and elbow–elbow morphologies). The appropriate robot morphology can be determined according to the terrain type. The lateral reachable body workspaces are derived analytically for different morphologies, which is useful for motion planning. Simulations and experiments are used to verify the design and analysis of the hexapod robot.