Naira Hovakimyan

Ayoosh Bansal, Hunmin Kim, Simon Yu et al.

Perception of obstacles remains a critical safety concern for autonomous vehicles. Real-world collisions have shown that the autonomy faults leading to fatal collisions originate from obstacle existence detection. Open source autonomous driving implementations show a perception pipeline with complex interdependent Deep Neural Networks. These networks are not fully verifiable, making them unsuitable for safety-critical tasks. In this work, we present a safety verification of an existing LiDAR based classical obstacle detection algorithm. We establish strict bounds on the capabilities of this obstacle detection algorithm. Given safety standards, such bounds allow for determining LiDAR sensor properties that would reliably satisfy the standards. Such analysis has as yet been unattainable for neural network based perception systems. We provide a rigorous analysis of the obstacle detection system with empirical results based on real-world sensor data.

Qianzhong Chen, Junheng Li, Sheng Cheng et al. · stanford

Bipedal locomotion control is essential for humanoid robots to navigate complex, human-centric environments. While optimization-based control designs are popular for integrating sophisticated models of humanoid robots, they often require labor-intensive manual tuning. In this work, we address the challenges of parameter selection in bipedal locomotion control using DiffTune, a model-based autotuning method that leverages differential programming for efficient parameter learning. A major difficulty lies in balancing model fidelity with differentiability. We address this difficulty using a low-fidelity model for differentiability, enhanced by a Ground Reaction Force-and-Moment Network (GRFM-Net) to capture discrepancies between MPC commands and actual control effects. We validate the parameters learned by DiffTune with GRFM-Net in hardware experiments, which demonstrates the parameters' optimality in a multi-objective setting compared with baseline parameters, reducing the total loss by up to 40.5$\%$ compared with the expert-tuned parameters. The results confirm the GRFM-Net's effectiveness in mitigating the sim-to-real gap, improving the transferability of simulation-learned parameters to real hardware.

Jing Wu, Ran Tao, Pan Zhao et al.

Nitrogen (N) management is critical to sustain soil fertility and crop production while minimizing the negative environmental impact, but is challenging to optimize. This paper proposes an intelligent N management system using deep reinforcement learning (RL) and crop simulations with Decision Support System for Agrotechnology Transfer (DSSAT). We first formulate the N management problem as an RL problem. We then train management policies with deep Q-network and soft actor-critic algorithms, and the Gym-DSSAT interface that allows for daily interactions between the simulated crop environment and RL agents. According to the experiments on the maize crop in both Iowa and Florida in the US, our RL-trained policies outperform previous empirical methods by achieving higher or similar yield while using less fertilizers

Suiyao Chen, Jing Wu, Naira Hovakimyan et al.

Representation learning stands as one of the critical machine learning techniques across various domains. Through the acquisition of high-quality features, pre-trained embeddings significantly reduce input space redundancy, benefiting downstream pattern recognition tasks such as classification, regression, or detection. Nonetheless, in the domain of tabular data, feature engineering and selection still heavily rely on manual intervention, leading to time-consuming processes and necessitating domain expertise. In response to this challenge, we introduce ReConTab, a deep automatic representation learning framework with regularized contrastive learning. Agnostic to any type of modeling task, ReConTab constructs an asymmetric autoencoder based on the same raw features from model inputs, producing low-dimensional representative embeddings. Specifically, regularization techniques are applied for raw feature selection. Meanwhile, ReConTab leverages contrastive learning to distill the most pertinent information for downstream tasks. Experiments conducted on extensive real-world datasets substantiate the framework's capacity to yield substantial and robust performance improvements. Furthermore, we empirically demonstrate that pre-trained embeddings can seamlessly integrate as easily adaptable features, enhancing the performance of various traditional methods such as XGBoost and Random Forest.

Ran Tao, Pan Zhao, Jing Wu et al.

Crop management, including nitrogen (N) fertilization and irrigation management, has a significant impact on the crop yield, economic profit, and the environment. Although management guidelines exist, it is challenging to find the optimal management practices given a specific planting environment and a crop. Previous work used reinforcement learning (RL) and crop simulators to solve the problem, but the trained policies either have limited performance or are not deployable in the real world. In this paper, we present an intelligent crop management system which optimizes the N fertilization and irrigation simultaneously via RL, imitation learning (IL), and crop simulations using the Decision Support System for Agrotechnology Transfer (DSSAT). We first use deep RL, in particular, deep Q-network, to train management policies that require all state information from the simulator as observations (denoted as full observation). We then invoke IL to train management policies that only need a limited amount of state information that can be readily obtained in the real world (denoted as partial observation) by mimicking the actions of the previously RL-trained policies under full observation. We conduct experiments on a case study using maize in Florida and compare trained policies with a maize management guideline in simulations. Our trained policies under both full and partial observations achieve better outcomes, resulting in a higher profit or a similar profit with a smaller environmental impact. Moreover, the partial-observation management policies are directly deployable in the real world as they use readily available information.

Jing Wu, David Pichler, Daniel Marley et al.

A key challenge for much of the machine learning work on remote sensing and earth observation data is the difficulty in acquiring large amounts of accurately labeled data. This is particularly true for semantic segmentation tasks, which are much less common in the remote sensing domain because of the incredible difficulty in collecting precise, accurate, pixel-level annotations at scale. Recent efforts have addressed these challenges both through the creation of supervised datasets as well as the application of self-supervised methods. We continue these efforts on both fronts. First, we generate and release an improved version of the Agriculture-Vision dataset (Chiu et al., 2020b) to include raw, full-field imagery for greater experimental flexibility. Second, we extend this dataset with the release of 3600 large, high-resolution (10cm/pixel), full-field, red-green-blue and near-infrared images for pre-training. Third, we incorporate the Pixel-to-Propagation Module Xie et al. (2021b) originally built on the SimCLR framework into the framework of MoCo-V2 Chen et al.(2020b). Finally, we demonstrate the usefulness of this data by benchmarking different contrastive learning approaches on both downstream classification and semantic segmentation tasks. We explore both CNN and Swin Transformer Liu et al. (2021a) architectures within different frameworks based on MoCo-V2. Together, these approaches enable us to better detect key agricultural patterns of interest across a field from aerial imagery so that farmers may be alerted to problematic areas in a timely fashion to inform their management decisions. Furthermore, the release of these datasets will support numerous avenues of research for computer vision in remote sensing for agriculture.

Yikun Cheng, Pan Zhao, Naira Hovakimyan

Safe reinforcement learning (RL) with assured satisfaction of hard state constraints during training has recently received a lot of attention. Safety filters, e.g., based on control barrier functions (CBFs), provide a promising way for safe RL via modifying the unsafe actions of an RL agent on the fly. Existing safety filter-based approaches typically involve learning of uncertain dynamics and quantifying the learned model error, which leads to conservative filters before a large amount of data is collected to learn a good model, thereby preventing efficient exploration. This paper presents a method for safe and efficient RL using disturbance observers (DOBs) and control barrier functions (CBFs). Unlike most existing safe RL methods that deal with hard state constraints, our method does not involve model learning, and leverages DOBs to accurately estimate the pointwise value of the uncertainty, which is then incorporated into a robust CBF condition to generate safe actions. The DOB-based CBF can be used as a safety filter with model-free RL algorithms by minimally modifying the actions of an RL agent whenever necessary to ensure safety throughout the learning process. Simulation results on a unicycle and a 2D quadrotor demonstrate that the proposed method outperforms a state-of-the-art safe RL algorithm using CBFs and Gaussian processes-based model learning, in terms of safety violation rate, and sample and computational efficiency.

Jing Wu, Jennifer Hobbs, Naira Hovakimyan

Contrastive learning models based on Siamese structure have demonstrated remarkable performance in self-supervised learning. Such a success of contrastive learning relies on two conditions, a sufficient number of positive pairs and adequate variations between them. If the conditions are not met, these frameworks will lack semantic contrast and be fragile on overfitting. To address these two issues, we propose Hallucinator that could efficiently generate additional positive samples for further contrast. The Hallucinator is differentiable and creates new data in the feature space. Thus, it is optimized directly with the pre-training task and introduces nearly negligible computation. Moreover, we reduce the mutual information of hallucinated pairs and smooth them through non-linear operations. This process helps avoid over-confident contrastive learning models during the training and achieves more transformation-invariant feature embeddings. Remarkably, we empirically prove that the proposed Hallucinator generalizes well to various contrastive learning models, including MoCoV1&V2, SimCLR and SimSiam. Under the linear classification protocol, a stable accuracy gain is achieved, ranging from 0.3% to 3.0% on CIFAR10&100, Tiny ImageNet, STL-10 and ImageNet. The improvement is also observed in transferring pre-train encoders to the downstream tasks, including object detection and segmentation.

Jing Wu, Naira Hovakimyan, Jennifer Hobbs

Classifying and segmenting patterns from a limited number of examples is a significant challenge in remote sensing and earth observation due to the difficulty in acquiring accurately labeled data in large quantities. Previous studies have shown that meta-learning, which involves episodic training on query and support sets, is a promising approach. However, there has been little attention paid to direct fine-tuning techniques. This paper repurposes contrastive learning as a pre-training method for few-shot learning for classification and semantic segmentation tasks. Specifically, we introduce a generator-based contrastive learning framework (GenCo) that pre-trains backbones and simultaneously explores variants of feature samples. In fine-tuning, the auxiliary generator can be used to enrich limited labeled data samples in feature space. We demonstrate the effectiveness of our method in improving few-shot learning performance on two key remote sensing datasets: Agriculture-Vision and EuroSAT. Empirically, our approach outperforms purely supervised training on the nearly 95,000 images in Agriculture-Vision for both classification and semantic segmentation tasks. Similarly, the proposed few-shot method achieves better results on the land-cover classification task on EuroSAT compared to the results obtained from fully supervised model training on the dataset.

Yite Wang, Jing Wu, Naira Hovakimyan et al.

Over the past few years, there has been growing interest in developing larger and deeper neural networks, including deep generative models like generative adversarial networks (GANs). However, GANs typically come with high computational complexity, leading researchers to explore methods for reducing the training and inference costs. One such approach gaining popularity in supervised learning is dynamic sparse training (DST), which maintains good performance while enjoying excellent training efficiency. Despite its potential benefits, applying DST to GANs presents challenges due to the adversarial nature of the training process. In this paper, we propose a novel metric called the balance ratio (BR) to study the balance between the sparse generator and discriminator. We also introduce a new method called balanced dynamic sparse training (ADAPT), which seeks to control the BR during GAN training to achieve a good trade-off between performance and computational cost. Our proposed method shows promising results on multiple datasets, demonstrating its effectiveness.

Wenbin Wan, Hunmin Kim, Naira Hovakimyan et al.

In this paper, an attack-resilient estimation algorithm is presented for linear discrete-time stochastic systems with state and input constraints. It is shown that the state estimation errors of the proposed estimation algorithm are practically exponentially stable.

Neng Wan, Dapeng Li, Naira Hovakimyan

A simplified approach is proposed to investigate the continuous-time and discrete-time complementary sensitivity Bode integrals (CSBIs) in this note. For continuous-time feedback systems with unbounded frequency domain, the CSBI weighted by $1/ω^2$ is considered, where this simplified method reveals a more explicit relationship between the value of CSBI and the structure of the open-loop transfer function. With a minor modification of this method, the CSBI of discrete-time system is derived, and illustrative examples are provided. Compared with the existing results on CSBI, neither Cauchy integral theorem nor Poisson integral formula are used throughout the analysis, and the analytic constraint on the integrand is removed.

Ayoosh Bansal, Hunmin Kim, Simon Yu et al.

Advances in deep learning have revolutionized cyber-physical applications, including the development of Autonomous Vehicles. However, real-world collisions involving autonomous control of vehicles have raised significant safety concerns regarding the use of Deep Neural Networks (DNN) in safety-critical tasks, particularly Perception. The inherent unverifiability of DNNs poses a key challenge in ensuring their safe and reliable operation. In this work, we propose Perception Simplex (PS), a fault-tolerant application architecture designed for obstacle detection and collision avoidance. We analyze an existing LiDAR-based classical obstacle detection algorithm to establish strict bounds on its capabilities and limitations. Such analysis and verification have not been possible for deep learning-based perception systems yet. By employing verifiable obstacle detection algorithms, PS identifies obstacle existence detection faults in the output of unverifiable DNN-based object detectors. When faults with potential collision risks are detected, appropriate corrective actions are initiated. Through extensive analysis and software-in-the-loop simulations, we demonstrate that PS provides predictable and deterministic fault tolerance against obstacle existence detection faults, establishing a robust safety guarantee.

Minhyuk Jang, Jungjin Lee, Astghik Hakobyan et al.

The extended Kalman filter (EKF) is a cornerstone of nonlinear state estimation, yet its performance is fundamentally limited by noise-model mismatch and linearization errors. We develop a residual-aware distributionally robust EKF that addresses both challenges within a unified Wasserstein distributionally robust state estimation framework. The key idea is to treat linearization residuals as uncertainty and absorb them into an effective uncertainty model captured by a stage-wise ambiguity set, enabling noise-model mismatch and approximation errors to be handled within a single formulation. This approach yields a computable effective radius along with deterministic upper bounds on the prior and posterior mean-squared errors of the true nonlinear estimation error. The resulting filter admits a tractable semidefinite programming reformulation while preserving the recursive structure of the classical EKF. Simulations on coordinated-turn target tracking and uncertainty-aware robot navigation demonstrate improved estimation accuracy and safety compared to standard EKF baselines under model mismatch and nonlinear effects.

Jinghan Yang, Hunmin Kim, Wenbin Wan et al.

Autonomous systems increasingly rely on machine learning techniques to transform high-dimensional raw inputs into predictions that are then used for decision-making and control. However, it is often easy to maliciously manipulate such inputs and, as a result, predictions. While effective techniques have been proposed to certify the robustness of predictions to adversarial input perturbations, such techniques have been disembodied from control systems that make downstream use of the predictions. We propose the first approach for composing robustness certification of predictions with respect to raw input perturbations with robust control to obtain certified robustness of control to adversarial input perturbations. We use a case study of adaptive vehicle control to illustrate our approach and show the value of the resulting end-to-end certificates through extensive experiments.

Astghik Hakobyan, Amaras Nazarians, Aditya Gahlawat et al.

Safe operation of autonomous systems requires robustness to both model uncertainty and uncertainty in the environment. We propose a hierarchical framework for stochastic nonlinear systems that integrates distributionally robust model predictive control (DR-MPC) with $\mathcal{L}_1$-adaptive control. The key idea is to use the $\mathcal{L}_1$ adaptive controller's online distributional certificates that bound the Wasserstein distance between nominal and true state distributions, thereby certifying the ambiguity sets used for planning without requiring distribution samples. Environment uncertainty is captured via data-driven ambiguity sets constructed from finite samples. These are incorporated into a DR-MPC planner enforcing distributionally robust chance constraints over a receding horizon. Using Wasserstein duality, the resulting problem admits tractable reformulations and a sample-based implementation. We show theoretically and via numerical experimentation that our framework ensures certifiable safety in the presence of simultaneous system and environment uncertainties.

Ran Tao, Pan Zhao, Ilya Kolmanovsky et al.

This paper introduces an uncertainty compensation-based robust adaptive model predictive control (MPC) framework for linear systems with nonlinear time-varying uncertainties. The framework integrates an L1 adaptive controller to compensate for the matched uncertainty and a robust feedback controller, designed using linear matrix inequalities, to mitigate the effect of unmatched uncertainty on target output channels. Uniform bounds on the errors between the system's states and control inputs and those of a nominal (i.e., uncertainty-free) system are derived. These error bounds are then used to tighten the actual system's state and input constraints, enabling the design of an MPC for the nominal system under these tightened constraints. Referred to as uncertainty compensation-based MPC (UC-MPC), this approach ensures constraint satisfaction while delivering enhanced performance compared to existing methods. Simulation results for a flight control example and a spacecraft landing on an asteroid demonstrate the effectiveness of the proposed framework.

Mikayel Aramyan, Anna Manucharyan, Lusine Poghosyan et al.

Coordinated missions involving Unmanned Aerial Vehicles (UAVs) in dynamic environments pose significant challenges in maintaining both coordination and agility. In this paper, relying on the cooperative path following framework and using a game-theoretic formulation, we introduce a novel and scalable approach in which each UAV acts autonomously in different mission conditions. This formulation naturally accommodates heterogeneous and time-varying objectives across the system. In our setting, each UAV optimizes a cost function that incorporates temporal and mission-specific constraints. The optimization is performed within a one-dimensional domain, significantly reducing the computational cost and enabling real-time application to complex and dynamic scenarios. The framework is distributed in structure, enabling global, system-wide coordination (a Nash equilibrium) by using only local information. For ideal systems, we prove the existence and the Nash equilibrium exhibits exponential convergence. Furthermore, we invoke model predictive control (MPC) for non-ideal scenarios. In particular, we propose a discrete-time optimization approach that tackles path-following errors and communication failures, ensuring reliable and agile performance in dynamic and uncertain environments. Simulation results demonstrate the effectiveness and agility of the approach in ensuring successful mission execution across diverse realistic scenarios.

Aditya Gahlawat, Ahmed Aboudonia, Sandeep Banik et al.

Imitation learning (IL) enables autonomous behavior by learning from expert demonstrations. While more sample-efficient than comparative alternatives like reinforcement learning, IL is sensitive to compounding errors induced by distribution shifts. There are two significant sources of distribution shifts when using IL-based feedback laws on systems: distribution shifts caused by policy error and distribution shifts due to exogenous disturbances and endogenous model errors due to lack of learning. Our previously developed approaches, Taylor Series Imitation Learning (TaSIL) and $\mathcal{L}_1$ -Distributionally Robust Adaptive Control (\ellonedrac), address the challenge of distribution shifts in complementary ways. While TaSIL offers robustness against policy error-induced distribution shifts, \ellonedrac offers robustness against distribution shifts due to aleatoric and epistemic uncertainties. To enable certifiable IL for learned and/or uncertain dynamical systems, we formulate \textit{Distributionally Robust Imitation Policy (DRIP)} architecture, a Layered Control Architecture (LCA) that integrates TaSIL and~\ellonedrac. By judiciously designing individual layer-centric input and output requirements, we show how we can guarantee certificates for the entire control pipeline. Our solution paves the path for designing fully certifiable autonomy pipelines, by integrating learning-based components, such as perception, with certifiable model-based decision-making through the proposed LCA approach.

Eshika Pathak, Ahmed Aboudonia, Sandeep Banik et al.

Dynamical system (DS)-based learning from demonstration (LfD) is a powerful tool for generating motion plans in the operation (`task') space of robotic systems. However, the realization of the generated motion plans is often compromised by a ''task-execution mismatch'', where unmodeled dynamics, persistent disturbances, and system latency cause the robot's actual task-space state to diverge from the desired motion trajectory. We propose a novel task-level robust control architecture, L1-augmented Dynamical Systems (L1-DS), that explicitly handles the task-execution mismatch in tracking a nominal motion plan generated by any DS-based LfD scheme. Our framework augments any DS-based LfD model with a nominal stabilizing controller and an L1 adaptive controller. Furthermore, we introduce a windowed Dynamic Time Warping (DTW)-based target selector, which enables the nominal stabilizing controller to handle temporal misalignment for improved phase-consistent tracking. We demonstrate the efficacy of our architecture on the LASA and IROS handwriting datasets.

Neng Wan, Dapeng Li, Naira Hovakimyan

Sensitivity of linear continuous-time control systems, subject to control and measurement noise, is analyzed by deriving the lower bounds of Bode-like integrals via an information-theoretic approach. Bode integrals of four different sensitivity-like functions are employed to gauge the control trade-offs. When the signals of the control system are stationary Gaussian, these four different Bode-like integrals can be represented as differences between mutual information rates. These mutual information rates and hence the corresponding Bode-like integrals are proven to be bounded below by the unstable poles and zeros of the plant model, if the signals of the control system are wide-sense stationary.

Jing Wu, Zhixin Lai, Suiyao Chen et al.

Crop management plays a crucial role in determining crop yield, economic profitability, and environmental sustainability. Despite the availability of management guidelines, optimizing these practices remains a complex and multifaceted challenge. In response, previous studies have explored using reinforcement learning with crop simulators, typically employing simple neural-network-based reinforcement learning (RL) agents. Building on this foundation, this paper introduces a more advanced intelligent crop management system. This system uniquely combines RL, a language model (LM), and crop simulations facilitated by the Decision Support System for Agrotechnology Transfer (DSSAT). We utilize deep RL, specifically a deep Q-network, to train management policies that process numerous state variables from the simulator as observations. A novel aspect of our approach is the conversion of these state variables into more informative language, facilitating the language model's capacity to understand states and explore optimal management practices. The empirical results reveal that the LM exhibits superior learning capabilities. Through simulation experiments with maize crops in Florida (US) and Zaragoza (Spain), the LM not only achieves state-of-the-art performance under various evaluation metrics but also demonstrates a remarkable improvement of over 49\% in economic profit, coupled with reduced environmental impact when compared to baseline methods. Our code is available at \url{https://github.com/jingwu6/LM_AG}.

Minkyung Kim, Henry Che, Bhargav Chandaka et al.

Accurate visual state estimation has been a central topic in robotics with a wide range of applications in robot navigation, autonomous driving, and autonomous flight. Recent advances in robot perception have led to significant improvements in the accuracy and robustness of state estimation, yet a fundamental challenge remains in how to quantify and calibrate its precision, i.e., how confident we are in an estimate and whether failures can be detected. This issue is particularly pronounced in visual-inertial odometry (VIO), where the heteroscedastic and multimodal nature of the problem makes uncertainty quantification especially difficult. This paper introduces MUSE (Multimodal Uncertainty Quantification of State Estimation), a novel real-time learning-based framework that leverages the strong and efficient sequential modeling capacity of Mamba to estimate localization uncertainty from multiple asynchronous sensor streams. Experiments on both public and in-house datasets demonstrate that MUSE achieves superior reliability and robustness compared to existing uncertainty quantification methods, and ablation studies justify the benefits of its key design choices.

Zhixin Lai, Jing Wu, Suiyao Chen et al.

In this study, we uncover the unexpected efficacy of residual-based large language models (LLMs) as part of encoders for biomedical imaging tasks, a domain traditionally devoid of language or textual data. The approach diverges from established methodologies by utilizing a frozen transformer block, extracted from pre-trained LLMs, as an innovative encoder layer for the direct processing of visual tokens. This strategy represents a significant departure from the standard multi-modal vision-language frameworks, which typically hinge on language-driven prompts and inputs. We found that these LLMs could boost performance across a spectrum of biomedical imaging applications, including both 2D and 3D visual classification tasks, serving as plug-and-play boosters. More interestingly, as a byproduct, we found that the proposed framework achieved superior performance, setting new state-of-the-art results on extensive, standardized datasets in MedMNIST-2D and 3D. Through this work, we aim to open new avenues for employing LLMs in biomedical imaging and enriching the understanding of their potential in this specialized domain.

Ayoosh Bansal, Mikael Yeghiazaryan, Artyom Khachatryan et al.

Autonomous systems increasingly rely on machine-learning (ML) components for safety-critical tasks such as perception and control in autonomous vehicles (AVs). While ML enables essential capabilities, it inevitably exhibits long-tail faults that make it unsuitable for safety-critical tasks. Runtime assurance (RTA) mitigates this issue by pairing ML components with verifiable safety monitors, e.g., Control Simplex and Perception Simplex architectures. However, the limited performance of safety monitors remains a major bottleneck. The Synergistic Simplex (SS) architecture improves system performance by enabling bidirectional integration between ML components and safety monitors while preserving formal safety guarantees. The key innovation here is allowing safety monitors to use ML outputs, which is typically prohibited in RTA systems. We formally derive conditions under which this integration preserves safety and demonstrate the performance benefits. We present the design, analysis, and evaluation of SS for AV obstacle detection.

Shengjie Liu, Jing Wu, Jingyuan Bao et al.

This paper describes an investigation of the robustness of large language models (LLMs) for retrieval augmented generation (RAG)-based summarization tasks. While LLMs provide summarization capabilities, their performance in complex, real-world scenarios remains under-explored. Our first contribution is LogicSumm, an innovative evaluation framework incorporating realistic scenarios to assess LLM robustness during RAG-based summarization. Based on limitations identified by LogiSumm, we then developed SummRAG, a comprehensive system to create training dialogues and fine-tune a model to enhance robustness within LogicSumm's scenarios. SummRAG is an example of our goal of defining structured methods to test the capabilities of an LLM, rather than addressing issues in a one-off fashion. Experimental results confirm the power of SummRAG, showcasing improved logical coherence and summarization quality. Data, corresponding model weights, and Python code are available online.

Yuliang Gu, Sheng Cheng, Naira Hovakimyan

Quadrotors are increasingly used in the evolving field of aerial robotics for their agility and mechanical simplicity. However, inherent uncertainties, such as aerodynamic effects coupled with quadrotors' operation in dynamically changing environments, pose significant challenges for traditional, nominal model-based control designs. We propose a multi-task meta-learning method called Encoder-Prototype-Decoder (EPD), which has the advantage of effectively balancing shared and distinctive representations across diverse training tasks. Subsequently, we integrate the EPD model into a model predictive control problem (Proto-MPC) to enhance the quadrotor's ability to adapt and operate across a spectrum of dynamically changing tasks with an efficient online implementation. We validate the proposed method in simulations, which demonstrates Proto-MPC's robust performance in trajectory tracking of a quadrotor being subject to static and spatially varying side winds.

Minjun Sung, Sambhu H. Karumanchi, Aditya Gahlawat et al.

We introduce $\mathcal{L}_1$-MBRL, a control-theoretic augmentation scheme for Model-Based Reinforcement Learning (MBRL) algorithms. Unlike model-free approaches, MBRL algorithms learn a model of the transition function using data and use it to design a control input. Our approach generates a series of approximate control-affine models of the learned transition function according to the proposed switching law. Using the approximate model, control input produced by the underlying MBRL is perturbed by the $\mathcal{L}_1$ adaptive control, which is designed to enhance the robustness of the system against uncertainties. Importantly, this approach is agnostic to the choice of MBRL algorithm, enabling the use of the scheme with various MBRL algorithms. MBRL algorithms with $\mathcal{L}_1$ augmentation exhibit enhanced performance and sample efficiency across multiple MuJoCo environments, outperforming the original MBRL algorithms, both with and without system noise.

Ani Vanyan, Alvard Barseghyan, Hakob Tamazyan et al.

Foundation models have advanced machine learning across various modalities, including images. Recently multiple teams trained foundation models specialized for remote sensing applications. This line of research is motivated by the distinct characteristics of remote sensing imagery, specific applications and types of robustness useful for satellite image analysis. In this work we systematically challenge the idea that specific foundation models are more useful than general-purpose vision foundation models, at least in the small scale. First, we design a simple benchmark that measures generalization of remote sensing models towards images with lower resolution for two downstream tasks. Second, we train iBOT, a self-supervised vision encoder, on MillionAID, an ImageNet-scale satellite imagery dataset, with several modifications specific to remote sensing. We show that none of those pretrained models bring consistent improvements upon general-purpose baselines at the ViT-B scale.

Yuliang Gu, Hongpeng Cao, Marco Caccamo et al.

Capturing latent variations ("contexts") is key to deploying reinforcement-learning (RL) agents beyond their training regime. We recast context-based RL as a dual inference-control problem and formally characterize two properties and their hierarchy: observation sufficiency (preserving all predictive information) and control sufficiency (retaining decision-making relevant information). Exploiting this dichotomy, we derive a contextual evidence lower bound(ELBO)-style objective that cleanly separates representation learning from policy learning and optimizes it with Bottlenecked Contextual Policy Optimization (BCPO), an algorithm that places a variational information-bottleneck encoder in front of any off-policy policy learner. On standard continuous-control benchmarks with shifting physical parameters, BCPO matches or surpasses other baselines while using fewer samples and retaining performance far outside the training regime. The framework unifies theory, diagnostics, and practice for context-based RL.

Yuliang Gu, Hongpeng Cao, Marco Caccamo et al.

The Cross-Entropy Method (CEM) is a widely adopted trajectory optimizer in model-based reinforcement learning (MBRL), but its unimodal sampling strategy often leads to premature convergence in multimodal landscapes. In this work, we propose Bregman Centroid Guided CEM ($\mathcal{BC}$-EvoCEM), a lightweight enhancement to ensemble CEM that leverages $\textit{Bregman centroids}$ for principled information aggregation and diversity control. $\textbf{$\mathcal{BC}$-EvoCEM}$ computes a performance-weighted Bregman centroid across CEM workers and updates the least contributing ones by sampling within a trust region around the centroid. Leveraging the duality between Bregman divergences and exponential family distributions, we show that $\textbf{$\mathcal{BC}$-EvoCEM}$ integrates seamlessly into standard CEM pipelines with negligible overhead. Empirical results on synthetic benchmarks, a cluttered navigation task, and full MBRL pipelines demonstrate that $\textbf{$\mathcal{BC}$-EvoCEM}$ enhances both convergence and solution quality, providing a simple yet effective upgrade for CEM.

Ashik E Rasul, Humaira Tasnim, Hyung-Jin Yoon et al.

Learning-based solutions have enabled incredible capabilities for autonomous systems. Autonomous vehicles, both aerial and ground, rely on DNN for various integral tasks, including perception. The efficacy of supervised learning solutions hinges on the quality of the training data. Discrepancies between training data and operating conditions result in faults that can lead to catastrophic incidents. However, collecting vast amounts of context-sensitive data, with broad coverage of possible operating environments, is prohibitively difficult. Synthetic data generation techniques for DNN allow for the easy exploration of diverse scenarios. However, synthetic data generation solutions for aerial vehicles are still lacking. This work presents a data augmentation framework for aerial vehicle's perception training, leveraging photorealistic simulation integrated with high-fidelity vehicle dynamics. Safe landing is a crucial challenge in the development of autonomous air taxis, therefore, landing maneuver is chosen as the focus of this work. With repeated simulations of landing in varying scenarios we assess the landing performance of the VTOL type UAV and gather valuable data. The landing performance is used as the objective function to optimize the DNN through retraining. Given the high computational cost of DNN retraining, we incorporated Bayesian Optimization in our framework that systematically explores the data augmentation parameter space to retrain the best-performing models. The framework allowed us to identify high-performing data augmentation parameters that are consistently effective across different landing scenarios. Utilizing the capabilities of this data augmentation framework, we obtained a robust perception model. The model consistently improved the perception-based landing success rate by at least 20% under different lighting and weather conditions.

Jing Wu, Zhixin Lai, Shengjie Liu et al.

Exploring the optimal management strategy for nitrogen and irrigation has a significant impact on crop yield, economic profit, and the environment. To tackle this optimization challenge, this paper introduces a deployable \textbf{CR}op Management system \textbf{O}ver all \textbf{P}ossible \textbf{S}tate availabilities (CROPS). CROPS employs a language model (LM) as a reinforcement learning (RL) agent to explore optimal management strategies within the Decision Support System for Agrotechnology Transfer (DSSAT) crop simulations. A distinguishing feature of this system is that the states used for decision-making are partially observed through random masking. Consequently, the RL agent is tasked with two primary objectives: optimizing management policies and inferring masked states. This approach significantly enhances the RL agent's robustness and adaptability across various real-world agricultural scenarios. Extensive experiments on maize crops in Florida, USA, and Zaragoza, Spain, validate the effectiveness of CROPS. Not only did CROPS achieve State-of-the-Art (SOTA) results across various evaluation metrics such as production, profit, and sustainability, but the trained management policies are also immediately deployable in over of ten millions of real-world contexts. Furthermore, the pre-trained policies possess a noise resilience property, which enables them to minimize potential sensor biases, ensuring robustness and generalizability. Finally, unlike previous methods, the strength of CROPS lies in its unified and elegant structure, which eliminates the need for pre-defined states or multi-stage training. These advancements highlight the potential of CROPS in revolutionizing agricultural practices.

Pan Zhao, Ziyao Guo, Yikun Cheng et al.

This paper presents an approach to trajectory-centric learning control based on contraction metrics and disturbance estimation for nonlinear systems subject to matched uncertainties. The approach uses deep neural networks to learn uncertain dynamics while still providing guarantees of transient tracking performance throughout the learning phase. Within the proposed approach, a disturbance estimation law is adopted to estimate the pointwise value of the uncertainty, with pre-computable estimation error bounds (EEBs). The learned dynamics, the estimated disturbances, and the EEBs are then incorporated in a robust Riemann energy condition to compute the control law that guarantees exponential convergence of actual trajectories to desired ones throughout the learning phase, even when the learned model is poor. On the other hand, with improved accuracy, the learned model can help improve the robustness of the tracking controller, e.g., against input delays, and can be incorporated to plan better trajectories with improved performance, e.g., lower energy consumption and shorter travel time.The proposed framework is validated on a planar quadrotor example.

Yikun Cheng, Pan Zhao, Manan Gandhi et al.

A reinforcement learning (RL) policy trained in a nominal environment could fail in a new/perturbed environment due to the existence of dynamic variations. Existing robust methods try to obtain a fixed policy for all envisioned dynamic variation scenarios through robust or adversarial training. These methods could lead to conservative performance due to emphasis on the worst case, and often involve tedious modifications to the training environment. We propose an approach to robustifying a pre-trained non-robust RL policy with $\mathcal{L}_1$ adaptive control. Leveraging the capability of an $\mathcal{L}_1$ control law in the fast estimation of and active compensation for dynamic variations, our approach can significantly improve the robustness of an RL policy trained in a standard (i.e., non-robust) way, either in a simulator or in the real world. Numerical experiments are provided to validate the efficacy of the proposed approach.

Saba Dadsetan, David Pichler, David Wilson et al.

Advances in remote sensing technology have led to the capture of massive amounts of data. Increased image resolution, more frequent revisit times, and additional spectral channels have created an explosion in the amount of data that is available to provide analyses and intelligence across domains, including agriculture. However, the processing of this data comes with a cost in terms of computation time and money, both of which must be considered when the goal of an algorithm is to provide real-time intelligence to improve efficiencies. Specifically, we seek to identify nutrient deficient areas from remotely sensed data to alert farmers to regions that require attention; detection of nutrient deficient areas is a key task in precision agriculture as farmers must quickly respond to struggling areas to protect their harvests. Past methods have focused on pixel-level classification (i.e. semantic segmentation) of the field to achieve these tasks, often using deep learning models with tens-of-millions of parameters. In contrast, we propose a much lighter graph-based method to perform node-based classification. We first use Simple Linear Iterative Cluster (SLIC) to produce superpixels across the field. Then, to perform segmentation across the non-Euclidean domain of superpixels, we leverage a Graph Convolutional Neural Network (GCN). This model has 4-orders-of-magnitude fewer parameters than a CNN model and trains in a matter of minutes.

Thiago Marinho, Massi Amrouche, Dusan Stipanovic et al.

Biological evidence shows that animals are capable of evading eminent collision without using depth information, relying solely on looming stimuli. In robotics, collision avoidance among uncooperative vehicles requires measurement of relative distance to the obstacle. Small, low-cost mobile robots and UAVs might be unable to carry distance measuring sensors, like LIDARS and depth cameras. We propose a control framework suitable for a unicycle-like vehicle moving in a 2D plane that achieves collision avoidance. The control strategy is inspired by the reaction of invertebrates to approaching obstacles, relying exclusively on line-of-sight (LOS) angle, LOS angle rate, and time-to-collision as feedback. Those quantities can readily be estimated from a monocular camera vision system onboard a mobile robot. The proposed avoidance law commands the heading angle to circumvent a moving obstacle with unknown position, while the velocity controller is left as a degree of freedom to accomplish other mission objectives. Theoretical guarantees are provided to show that minimum separation between the vehicle and the obstacle is attained regardless of the exogenous tracking controller.

Gabriel Barsi Haberfeld, Aditya Gahlawat, Naira Hovakimyan

Demand for fast and economical parcel deliveries in urban environments has risen considerably in recent years. A framework envisions efficient last-mile delivery in urban environments by leveraging a network of ride-sharing vehicles, where Unmanned Aerial Systems (UASs) drop packages on said vehicles, which then cover the majority of the distance before final aerial delivery. Notably, we consider the problem of planning a rendezvous path for the UAS to reach a human driver, who may choose between N possible paths and has uncertain behavior, while meeting strict safety constraints. The long planning horizon and safety constraints require robust heuristics that combine learning and optimal control using Gaussian Process Regression, sampling-based optimization, and Model Predictive Control. The resulting algorithm is computationally efficient and shown to be effective in a variety of qualitative scenarios.

Hyung-Jin Yoon, Hunmin Kim, Kripash Shrestha et al.

Path planning over spatiotemporal models can be applied to a variety of applications such as UAVs searching for spreading wildfire in mountains or network of balloons in time-varying atmosphere deployed for inexpensive internet service. A notable aspect in such applications is the dynamically changing environment. However, path planning algorithms often assume static environments and only consider the vehicle's dynamics exploring the environment. We present a spatiotemporal model that uses a cross-correlation operator to consider spatiotemporal dependence. Also, we present an adaptive state estimator for path planning. Since the state estimation depends on the vehicle's path, the path planning needs to consider the trade-off between exploration and exploitation. We use a high-level decision-maker to choose an explorative path or an exploitative path. The overall proposed framework consists of an adaptive state estimator, a short-term path planner, and a high-level decision-maker. We tested the framework with a spatiotemporal model simulation where the state of each grid transits from normal, latent, and fire state. For the mission objective of visiting the grids with fire, the proposed framework outperformed the random walk (baseline) and the single-minded exploitation (or exploration) path.

Neng Wan, Aditya Gahlawat, Naira Hovakimyan et al.

Distributed algorithms for both discrete-time and continuous-time linearly solvable optimal control (LSOC) problems of networked multi-agent systems (MASs) are investigated in this paper. A distributed framework is proposed to partition the optimal control problem of a networked MAS into several local optimal control problems in factorial subsystems, such that each (central) agent behaves optimally to minimize the joint cost function of a subsystem that comprises a central agent and its neighboring agents, and the local control actions (policies) only rely on the knowledge of local observations. Under this framework, we not only preserve the correlations between neighboring agents, but moderate the communication and computational complexities by decentralizing the sampling and computational processes over the network. For discrete-time systems modeled by Markov decision processes, the joint Bellman equation of each subsystem is transformed into a system of linear equations and solved using parallel programming. For continuous-time systems modeled by Itô diffusion processes, the joint optimality equation of each subsystem is converted into a linear partial differential equation, whose solution is approximated by a path integral formulation and a sample-efficient relative entropy policy search algorithm, respectively. The learned control policies are generalized to solve the unlearned tasks by resorting to the compositionality principle, and illustrative examples of cooperative UAV teams are provided to verify the effectiveness and advantages of these algorithms.

Jennifer Hobbs, Ivan Dozier, Naira Hovakimyan

"No-till" and cover cropping are often identified as the leading simple, best management practices for carbon sequestration in agriculture. However, the root of the problem is more complex, with the potential benefits of these approaches depending on numerous factors including a field's soil type(s), topography, and management history. Instead of using computer vision approaches to simply classify a field a still vs. no-till, we instead seek to identify the degree of residue coverage across afield through a probabilistic deep learning segmentation approach to enable more accurate analysis of carbon holding potential and realization. This approach will not only provide more precise insights into currently implemented practices, but also enable a more accurate identification process of fields with the greatest potential for adopting new practices to significantly impact carbon sequestration in agriculture.

Saba Dadsetan, Gisele Rose, Naira Hovakimyan et al.

Early, precise detection of nutrient deficiency stress (NDS) has key economic as well as environmental impact; precision application of chemicals in place of blanket application reduces operational costs for the growers while reducing the amount of chemicals which may enter the environment unnecessarily. Furthermore, earlier treatment reduces the amount of loss and therefore boosts crop production during a given season. With this in mind, we collect sequences of high-resolution aerial imagery and construct semantic segmentation models to detect and predict NDS across the field. Our work sits at the intersection of agriculture, remote sensing, and modern computer vision and deep learning. First, we establish a baseline for full-field detection of NDS and quantify the impact of pretraining, backbone architecture, input representation, and sampling strategy. We then quantify the amount of information available at different points in the season by building a single-timestamp model based on a UNet. Next, we construct our proposed spatiotemporal architecture, which combines a UNet with a convolutional LSTM layer, to accurately detect regions of the field showing NDS; this approach has an impressive IOU score of 0.53. Finally, we show that this architecture can be trained to predict regions of the field which are expected to show NDS in a later flight -- potentially more than three weeks in the future -- maintaining an IOU score of 0.47-0.51 depending on how far in advance the prediction is made. We will also release a dataset which we believe will benefit the computer vision, remote sensing, as well as agriculture fields. This work contributes to the recent developments in deep learning for remote sensing and agriculture, while addressing a key social challenge with implications for economics and sustainability.

Neng Wan, Aditya Gahlawat, Naira Hovakimyan et al.

A distributed stochastic optimal control solution is presented for cooperative multi-agent systems. The network of agents is partitioned into multiple factorial subsystems, each of which consists of a central agent and neighboring agents. Local control actions that rely only on agents' local observations are designed to optimize the joint cost functions of subsystems. When solving for the local control actions, the joint optimality equation for each subsystem is cast as a linear partial differential equation and solved using the Feynman-Kac formula. The solution and the optimal control action are then formulated as path integrals and approximated by a Monte-Carlo method. Numerical verification is provided through a simulation example consisting of a team of cooperative UAVs.

Lin Song, Neng Wan, Aditya Gahlawat et al.

In this paper, we discuss the methodology of generalizing the optimal control law from learned component tasks to unlearned composite tasks on Multi-Agent Systems (MASs), by using the linearity composition principle of linearly solvable optimal control (LSOC) problems. The proposed approach achieves both the compositionality and optimality of control actions simultaneously within the cooperative MAS framework in both discrete- and continuous-time in a sample-efficient manner, which reduces the burden of re-computation of the optimal control solutions for the new task on the MASs. We investigate the application of the proposed approach on the MAS with coordination between agents. The experiments show feasible results in investigated scenarios, including both discrete and continuous dynamical systems for task generalization without resampling.

Neng Wan, Dapeng Li, Naira Hovakimyan

This paper introduces the $f$-divergence variational inference ($f$-VI) that generalizes variational inference to all $f$-divergences. Initiated from minimizing a crafty surrogate $f$-divergence that shares the statistical consistency with the $f$-divergence, the $f$-VI framework not only unifies a number of existing VI methods, e.g. Kullback-Leibler VI, Rényi's $α$-VI, and $χ$-VI, but offers a standardized toolkit for VI subject to arbitrary divergences from $f$-divergence family. A general $f$-variational bound is derived and provides a sandwich estimate of marginal likelihood (or evidence). The development of the $f$-VI unfolds with a stochastic optimization scheme that utilizes the reparameterization trick, importance weighting and Monte Carlo approximation; a mean-field approximation scheme that generalizes the well-known coordinate ascent variational inference (CAVI) is also proposed for $f$-VI. Empirical examples, including variational autoencoders and Bayesian neural networks, are provided to demonstrate the effectiveness and the wide applicability of $f$-VI.

Aditya Gahlawat, Arun Lakshmanan, Lin Song et al.

We present $\mathcal{CL}_1$-$\mathcal{GP}$, a control framework that enables safe simultaneous learning and control for systems subject to uncertainties. The two main constituents are contraction theory-based $\mathcal{L}_1$ ($\mathcal{CL}_1$) control and Bayesian learning in the form of Gaussian process (GP) regression. The $\mathcal{CL}_1$ controller ensures that control objectives are met while providing safety certificates. Furthermore, $\mathcal{CL}_1$-$\mathcal{GP}$ incorporates any available data into a GP model of uncertainties, which improves performance and enables the motion planner to achieve optimality safely. This way, the safe operation of the system is always guaranteed, even during the learning transients. We provide a few illustrative examples for the safe learning and control of planar quadrotor systems in a variety of environments.

Mang Tik Chiu, Xingqian Xu, Kai Wang et al.

The first Agriculture-Vision Challenge aims to encourage research in developing novel and effective algorithms for agricultural pattern recognition from aerial images, especially for the semantic segmentation task associated with our challenge dataset. Around 57 participating teams from various countries compete to achieve state-of-the-art in aerial agriculture semantic segmentation. The Agriculture-Vision Challenge Dataset was employed, which comprises of 21,061 aerial and multi-spectral farmland images. This paper provides a summary of notable methods and results in the challenge. Our submission server and leaderboard will continue to open for researchers that are interested in this challenge dataset and task; the link can be found here.

Arun Lakshmanan, Aditya Gahlawat, Naira Hovakimyan

Autonomous robots that are capable of operating safely in the presence of imperfect model knowledge or external disturbances are vital in safety-critical applications. In this paper, we present a planner-agnostic framework to design and certify safe tubes around desired trajectories that the robot is always guaranteed to remain inside of. By leveraging recent results in contraction analysis and $\mathcal{L}_1$-adaptive control we synthesize an architecture that induces safe tubes for nonlinear systems with state and time-varying uncertainties. We demonstrate with a few illustrative examples how contraction theory-based $\mathcal{L}_1$-adaptive control can be used in conjunction with traditional motion planning algorithms to obtain provably safe trajectories.

Andrew Patterson, Aditya Gahlawat, Naira Hovakimyan

Autonomous agents must be able to safely interact with other vehicles to integrate into urban environments. The safety of these agents is dependent on their ability to predict collisions with other vehicles' future trajectories for replanning and collision avoidance. The information needed to predict collisions can be learned from previously observed vehicle trajectories in a specific environment, generating a traffic model. The learned traffic model can then be incorporated as prior knowledge into any trajectory estimation method being used in this environment. This work presents a Gaussian process based probabilistic traffic model that is used to quantify vehicle behaviors in an intersection. The Gaussian process model provides estimates for the average vehicle trajectory, while also capturing the variance between the different paths a vehicle may take in the intersection. The method is demonstrated on a set of time-series position trajectories. These trajectories are reconstructed by removing object recognition errors and missed frames that may occur due to data source processing. To create the intersection traffic model, the reconstructed trajectories are clustered based on their source and destination lanes. For each cluster, a Gaussian process model is created to capture the average behavior and the variance of the cluster. To show the applicability of the Gaussian model, the test trajectories are classified with only partial observations. Performance is quantified by the number of observations required to correctly classify the vehicle trajectory. Both the intersection traffic modeling computations and the classification procedure are timed. These times are presented as results and demonstrate that the model can be constructed in a reasonable amount of time and the classification procedure can be used for online applications.

Mang Tik Chiu, Xingqian Xu, Yunchao Wei et al.

The success of deep learning in visual recognition tasks has driven advancements in multiple fields of research. Particularly, increasing attention has been drawn towards its application in agriculture. Nevertheless, while visual pattern recognition on farmlands carries enormous economic values, little progress has been made to merge computer vision and crop sciences due to the lack of suitable agricultural image datasets. Meanwhile, problems in agriculture also pose new challenges in computer vision. For example, semantic segmentation of aerial farmland images requires inference over extremely large-size images with extreme annotation sparsity. These challenges are not present in most of the common object datasets, and we show that they are more challenging than many other aerial image datasets. To encourage research in computer vision for agriculture, we present Agriculture-Vision: a large-scale aerial farmland image dataset for semantic segmentation of agricultural patterns. We collected 94,986 high-quality aerial images from 3,432 farmlands across the US, where each image consists of RGB and Near-infrared (NIR) channels with resolution as high as 10 cm per pixel. We annotate nine types of field anomaly patterns that are most important to farmers. As a pilot study of aerial agricultural semantic segmentation, we perform comprehensive experiments using popular semantic segmentation models; we also propose an effective model designed for aerial agricultural pattern recognition. Our experiments demonstrate several challenges Agriculture-Vision poses to both the computer vision and agriculture communities. Future versions of this dataset will include even more aerial images, anomaly patterns and image channels. More information at https://www.agriculture-vision.com.