Modeling the quantitative relationship between target components and measured spectral information is an essential part of laser-induced breakdown spectroscopy (LIBS) analysis. However, many ...traditional multivariate analysis algorithms must reduce the spectral dimension or extract the characteristic spectral lines in advance, which may result in information loss and reduced accuracy. Indeed, improving the precision and interpretability of LIBS quantitative analysis is a critical challenge in Mars exploration. To solve this problem, this paper proposes an end-to-end lightweight quantitative modeling framework based on ensemble convolutional neural networks (ECNNs). This method eliminates the need for dimensionality reduction of the raw spectrum along with other pre-processing operations. We used the ChemCam calibration dataset as an example to verify the effectiveness of the proposed approach. Compared with partial least squares regression (a linear method) and extreme learning machine (a nonlinear method), our proposed method resulted in a lower root-mean-square error for major element prediction (54% and 73% lower, respectively) and was more stable. We also delved into the internal learning mechanism of the deep CNN model to understand how it hierarchically extracts spectral information features. The experimental results demonstrate that the easy-to-use ECNN-based regression model achieves excellent prediction performance while maintaining interpretability.
Reconfigurability in versatile systems of modular robots is achieved by changing the morphology of the overall structure as well as by connecting and disconnecting modules. Recurrent connectivity ...changes can cause misalignment that leads to mechanical failure of the system. This paper presents a new approach to reconfiguration, inspired by the art of origami, that eliminates connectivity changes during transformation. Our method consists of an energy-optimal reconfiguration planner that generates an initial 2D assembly pattern and an actuation sequence of the modular units, both resulting in minimum energy consumption. The algorithmic framework includes two approaches, an automatic modeling algorithm as well as a heuristic algorithm. We further demonstrate the effectiveness of our method by applying the algorithms to Mori, a modular origami robot, in simulation. Our results show that the heuristic algorithm yields reconfiguration schemes with high quality, compared with the automatic modeling algorithm, simultaneously saving a considerable amount of computational time and effort.
An accurate, nondestructive, and low-cost measurement system was developed using a portable near-infrared (NIR) spectrometer (DLP NIRscan Nano), a Raspberry Pi board, a display, a lithium battery, ...and a self-made three-dimensional printed shell. NIR data were collected through two measurement modes (column and Hadamard transform) based on digital light processing. With this equipment, detection models of soluble solid content (SSC) and firmness, essential quality indicators of the fruit, were established via quantitative analysis using chemometrics and a hybrid wavelength selection strategy. The SSC and firmness prediction model established through the combination of the synergy interval partial least squares and genetic algorithm (Si-GA-PLS) showed higher prediction accuracy, with coefficient of determination of prediction (RP2) values of 0.9406 and 0.9119, respectively, and root-mean-square error of prediction (RMSEP) values of 0.1655 and 5.5003, respectively. A comparison of the model performance of different monochromator principles was also explored; they were found to be non-statistically significant differences from one another. Finally, data fusion was used to improve prediction ability. The results obtained by mid-level data fusion presented a better performance than using models based on one technique. Overall, the developed novel handheld detector exhibits potential for smart software applications with high accuracy.
•Si-GA-PLS is a good method to deal with the near-infrared reflectance spectrum.•From a low-cost NIR evaluation module, a high-performance portable NIR spectrometer was proposed.•Compact hardware size and Cloud computing data processing software for on-site detection.•NIR spectroscopy has a broad application prospect in food safety monitoring in the future.•NIR spectroscopy from two portable systems was fused to improve prediction performance.
Near-infrared spectroscopy (NIRS) is one of the most promising technique for nondestructive and rapid detection of fruit's soluble solid content (SSC). However, when using NIRS to assess the SSC of ...fruit, a strategy for individually modeling different fruit cultivars is usually required, while the maintenance and upgrading of the models are time-consuming and laborious. To cope with these problems, this study aimed to explore the feasibility of developing a universal model to predict SSC for thin-skinned fruits with similar physicochemical properties. A progressive hybrid variable selection strategy was used to establish the universal model to decrease the complexity of the modeling and ultimately increase the model accuracy. First, the characteristic wavebands of four cultivars were chosen by synergy interval partial least squares (Si-PLS). Next, the wavelength point selection method was employed to filter the uninformative wavelengths and then mapped to an L1 regularization optimization task with constraints. Finally, the effective variables were further identified by simulated annealing (SA) and genetic algorithm (GA), separating them from the remaining variables. The coefficients of determination of calibration (RC2) and prediction (RP2) were both 0.93, and the root mean square error of calibration (RMSECV) and prediction (RMSEP) were 0.62 °Brix and 0.60 °Brix, respectively, for the Si-L1-UVE-GA model. A new external sample set was used for external prediction, and the RMSEP and Rp2 for the multi-cultivars model were 0.73 °Brix and 0.90, respectively. The results showed superiority of developing a universal model for predicting SSC of different species by using wavelength-limited portable NIR equipment.
Display omitted
•A universal NIR model for SSC testing from four cultivars was developed.•SG smoothing and MSC pre-processing methods were useful in developing a universal model.•L1 regularization method is adopted to find the best combination of input variables.•The proposed hybrid strategy reduced the complexity of the universal model.•The portable NIR spectrometer successfully predicted SSC in several species of fruits.
Abstract Since rocks may collide with the rover or wear tires during the exploration mission of the Mars probe, and may contain rich geological information, identifying rocks in the scene is crucial ...for the navigation and obstacle avoidance of the Mars probe. Additionally, since the communication between the Mars rover and the earth is often intermittent and delayed during missions, it needs a certain degree of autonomy. Deep learning technologies such as semantic segmentation and target detection can meet this requirement to a certain extent, which facilitates the enhancement of safety and efficiency for the Mars rover. Rock segmentation is to divide the pixels of the rock from the image. However, the texture of the rock is often close to the texture of the surrounding sand, and some parts may be covered, so it is difficult to identify it correctly. To this end, this paper proposed RSU-Net (Rock Segmentation U-Net) and RSU-Net-L, which combine the SENet attention mechanism, and the latter achieves higher computational efficiency and inference speed by compressing the number of channels on the basis of the former. In addition, this paper established a dataset, MarsRock, for Mars rock segmentation to help the Mars rover for visual navigation. Its images come from “Tianwen-1”, which contains 1194 images, and each image has a corresponding rock label. And our experiments on the MarsRock dataset show that RSU-Net can achieve 99.07% accuracy and 67.71% F1-score. RSU-Net-L can achieve 98.99% accuracy and 66.67% F1-score while diminishing the number of parameter count by 43.7% and the number of FLOPs by 43.6%, while the FPS can reach 12.01 on a single RTX6000-24GB GPU.
Abstract Although deep reinforcement learning (DRL) has been widely used in robotic mapless navigation tasks, most of the current research focuses on structured environments such as indoor or maze ...scenes, and little is targeted on outdoor environments. Unlike indoor environments and mazes, outdoor fields tend to be unstructured with complex landforms and sparse rewards for robotic navigation. The performance of most DRL-based strategies is directly affected by the design of the reward function, which greatly deteriorates its generalizability to outdoor environments. To this end, here we propose a two-stage learning paradigm based on skill discovery and hierarchical reinforcement learning (HRL) to cope with this challenge. Specifically, we implement skill discovery through a pre-training stage to acquire diverse skills with terrain-adaptive exploration strategies; then we select multiple skills using HRL to cope with more complex scenarios. We carry out the robotic multi-terrain traverse task based on a high-fidelity robotic simulation platform, Webots, and implement extensive comparative experiments and ablation studies to demonstrate the effectiveness of our approach.
Abstract
With high demand of automatic development of four-wheel alignment technology, this paper explores the core technology of automatic detection of four-wheel alignment instrument: intelligent ...identification and processing of rim target. According to the working process of 3D vision four-wheel aligner, starting from the target positioning, this paper proposes a method based on matching idea to realize the target recognition and positioning; Then the target area image is segmented to obtain the center coordinates on it, which provides a basis for the subsequent calculation of four-wheel positioning parameters. Simulation results demonstrate the effectiveness of our method.
Martian rock segmentation aims to separate rock pixels from background, which plays a crucial role in downstream tasks, such as traversing and geologic analysis by Mars rovers. The U-Nets have ...achieved certain results in rock segmentation. However, due to the inherent locality of convolution operations, U-Nets are inadequate in modeling global context and long-range spatial dependencies. Although emerging Transformers can solve this, they suffer from difficulties in extracting and retaining sufficient low-level local information. These shortcomings limit the performance of existing networks for Martian rocks that are variable in shape, size, texture and color. Therefore, we propose RockFormer, the first U-shaped Transformer framework for Mars rock segmentation, consisting of a hierarchical encoder-decoder architecture with a Feature Refining Module (FRM) connected between them. Specifically, the encoder hierarchically generates multi-scale features using an improved vision Transformer (improved-ViT), where both abundant local information and long-range contexts are exploited. The FRM removes less representative features and captures global dependencies between multi-scale features, improving RockFormer's robustness to Martian rocks with diverse appearances. The decoder is responsible for aggregating these features for pixel-wise rock prediction. For evaluation, we establish two Mars rock datasets, including both real and synthesized images. One is MarsData-V2, an extension of our previously published MarsData collected from real Mars rocks. The other is SynMars, a synthetic dataset sequentially photographed from a virtual terrain built referring to the TianWen-1 dataset. Extensive experiments on the two datasets show the superiority of RockFormer for Martian rock segmentation, achieving state-of-the-art performance with decent computational simplicity.
Laser-induced breakdown spectroscopy (LIBS) is a useful tool for the
in situ
detection of material components on Mars. This technique can directly acquire key information, such as the types and ...abundance of elements in the surface rocks and soil of Mars, with submillimeter spatial precision. Because the LIBS signal is prone to interference from the sample matrix and self-absorption effects, as well as other factors, the accuracy of its qualitative and quantitative analyses can be diminished without proper treatment. High-resolution LIBS generates hundreds or thousands of variables, causing low efficiency or high probability of overfitting in wavelength selection algorithms. Chemometric methods can be used to overcome the effects of these factors, extract useful information from the complex spectral data, and establish more robust analytical models. In this paper, model population analysis (MPA) combined with a range of variable selection methods is used to analyze the standard ground samples of the ChemCam team quantitatively. The analytical performance of LIBS is improved by optimizing and improving the algorithms. The results show that the two-step hybrid modeling method based on MPA has excellent prediction performance. Compared with other single or hybrid models, our method enables extensive use of the strengths of the models used and compensates for their weaknesses when handling high-dimensional datasets. The model established in this paper may contribute to the quantitative analysis of LIBS spectral data in future Mars exploration missions.
Component prediction models for laser induced breakdown spectroscopy data of ChemCam are created using a hybrid variable selection strategy.
Semantic segmentation of Martian terrain is crucial for the route planning and autonomous navigation of rovers on Mars. However, existing methods are restricted to structured or semi-structured ...scenes, performing poorly on Mars that is a completely unstructured environment. Therefore, we propose a novel hybrid attention semantic segmentation (HASS) network, which contains a global intra-class attention branch, a local inter-class attention branch and a representation merging module. Specifically, the global attention branch draws the consistencies of all homogeneous pixels in the whole image, and the local attention branch models the relationships between specific heterogeneous pixels with the supervision of elaborately designed loss function. The merging module aggregates the contexts from the two branches for the final segmentation. Furthermore, we establish a panorama semantic segmentation dataset of Martian landforms, named MarsScapes, which provides fine-grained annotations for eight semantic categories. Extensive experiments on our MarsScapes and the public AI4Mars datasets show the superiority of the proposed method.
•We design a hybrid attention semantic segmentation method with a dual-branch network.•We establish a panorama dataset of Martian landforms with detailed annotations.•We demonstrate HASS outperforms existing approaches through extensive experiments.