Xiangyu Zhang

This paper explores the emerging synergy between control theory and machine learning in robotics, focusing on methods that combine model-based strategies with data-driven adaptation. The authors highlight how classical techniques, such as model predictive control and adaptive control, are being enhanced by reinforcement learning, imitation learning, and neural models to address challenges in complex, uncertain environments. Emphasis is placed on real-world platforms (e.g., legged systems, aerial robots, and manipulators) with special attention to advanced domains such as multi-agent systems and coordination. The authors, in addition, establish a taxonomy to categorize these hybrid approaches as “learning-for-control”, “control-for-learning”, or “co-designed architectures”. This paper also reflects upon key open problems, including sim-to-real transfer, safety, and the need for verifiable learning-based controllers, all facets that help to outline a roadmap for future research.

Background: Although genomic technologies are increasingly used in dementia risk assessment, diagnosis and treatment, the impact of genomic burden on dementia caregiving remains largely unexamined.Purpose: This research investigated the impact of genomic burden on dementia caregiving and identified gaps to inform dementia care practice and policy.Methods: Three electronic databases, PubMed, APA PsycInfo, and CINAHL, were searched for peer-reviewed studies examining caregiver outcomes related to genomic burden in persons living with dementia (PLWD), published by December 2025. Two reviewers independently screened articles, resolving discrepancies by consensus. Data were extracted and thematically analyzed.Results: Of 741 studies identified, 91 underwent full-text screening. 8 studies met inclusion criteria. Genomic factors investigated included APOE (n = 6), LDLR (n = 1), ACE (n = 1), PSEN (n = 2), and MAO-A (n = 2) variants. Care recipients’ diagnoses ranged from early-onset Alzheimer’s disease to Alzheimer’s disease and related dementias and represented a wide range of disease severity. Caregivers were primarily family members. Caregiving outcomes included caregiving burden and caregiving distress.Conclusion: Though there is limited evidence to determine the impact of genomic factors on caregiving outcomes, existing studies suggest that care recipients' genetic factors influence caregiver burden indirectly through behavioral and psychological symptoms of dementia. Longitudinal studies are needed to assess the long-term impact of and the specific mechanisms by which genomic burden impacts dementia caregiving. Intervention studies integrating genetic counseling in dementia care models could clarify clinical utility as well as the psychosocial effects of genomic risk disclosure on the caregiving process.

Background: Although genomic technologies are increasingly used in dementia risk assessment, diagnosis and treatment, the impact of genomic burden in persons living with dementia (PWLD) on caregiving remains largely unexamined. Purpose: We investigated the impact of genomic burden on dementia caregiving and identified gaps to inform future practice and policy on dementia care. Methods: We searched PubMed, APA PsycInfo, and CINAHL through April 2025 for peer-reviewed studies examining caregiver outcomes related to genomic burden in PLWD. Two reviewers independently screened articles, resolving discrepancies by consensus. Data were extracted and thematically analyzed. Results: Of 708 studies identified, 82 underwent full-text screening. 5 studies that met inclusion criteria were conducted in the UK (n = 1) and Brazil (n = 4). Investigated genomic factors included APOE (n = 4), LDLD (n = 1), ACE (n = 1), and MAO-A (n = 1) variants. Care recipients, mainly with late-onset Alzheimer’s disease, represented a range of dementia severity from probable to severe. Caregivers were primarily family members. Caregiving outcomes included caregiving burden and caregiving distress. Conclusion: Available evidence to determine the impact of genomic burden in PWLD on caregiving outcomes is limited, but they point to the mediating role of neuropsychiatric symptoms. Longitudinal studies are needed to assess how genomic burden influences caregiving burden. Intervention studies integrating genomic information and/or genetic counseling in dementia care models could clarify clinical utility as well as the psychosocial effects of genomic risk disclosure on the caregiving process.

In this article, a robust explicit model predictive control (EMPC) flight scheme is investigated for a quadrotor. MPC is widely recognized for its control effectiveness, but the computational complexity involved in solving online optimization problems, particularly when applied to fast systems, poses a significant challenge. To enable real-time MPC implementation on quadrotor systems, we propose a novel dual-layer control architecture integrating EMPC, strategically relocating the computationally intensive optimization process to offline computation. The outer loop computes reference roll and pitch angles, while the inner loop employs an EMPC framework to achieve fast attitude tracking considering state and actuator constraints. Moreover, integral sliding mode control (ISMC) is integrated to mitigate the effects of uncertainties, such as unbalanced payloads. The recursive feasibility is guaranteed for the proposed flight control method if the initial states are in the feasibility set, and the Lyapunov stability analysis is conducted. In addition, we develop a polynomial trajectory planning algorithm for the quadrotor in (3-D) space. We employ our previous result, the bidirectional guidance informed trees (BIGIT*) algorithm, to obtain a sequence of collision-free waypoints, and utilize the minimum-snap technique to generate a smooth path. Finally, experimental results demonstrate the effectiveness of the proposed methods.

This study explores the use of a robust Explicit Model Predictive Control (EMPC) scheme for an Autonomous Surface Vehicle (ASV). The aim is to manage computational complexity related to online optimization, particularly for fast-response ASVs which are equipped with micro hardware with limited computational capabilities. The optimization process is performed offline a priori to a given mission, and the proposed EMPC scheme is shown to ensure stability via Lyapunov analysis. The authors also investigate the way in which the resulting ASV trajectories, especially taking into account polynomial trajecto-ries, affect the total ASV control effort. This qualitative compari-son of trajectories and respective control efforts is the basis from which ASV control performance is analyzed. Numerical analysis confirms the effectiveness of the proposed EMPC-based control strategies. Additionally, it reveals how energy consumption varies with different polynomial trajectories of varying degrees.

In this paper, a robust explicit model predictive control (EMPC) flight scheme is investigated for a quadrotor. MPC is widely recognized for its effectiveness in control, but its computational complexity for solving the online optimization problem, particularly for fast systems, poses a challenge. To address this, we introduce an innovative inner-outer loop control structure that incorporates EMPC which can transfer the overall optimization process offline. In the outer loop, we calculate reference roll and pitch angles, while the inner loop employs EMPC for rapid and optimized angle tracking within practical constraints. Additionally, integral sliding mode control (ISMC) is integrated to mitigate the effects of unbalanced pay-loads. The recursive feasibility is guaranteed for the proposed flight control method if the initial states are in the feasibility set, and the Lyapunov stability analysis is conducted. Finally, experiments show the efficacy of the proposed control strategies.

Obfuscation is a prevalent security technique which transforms syntactic representation of a program to a complicated form, but still keeps program semantics unchanged. So far, developers heavily rely on obfuscation to harden their products and reduce the risk of adversarial reverse engineering. However, despite its spectacular progress, one crucial hurdle is that each of existing obfuscation method is designed specifically for obfuscating one program feature (e.g., identifier name, control flow), so an effective obfuscation scheme usually composes a considerable amount of different obfuscation methods. Therefore, one primary challenge lies in identifying effective combinations of obfuscation methods. In this research, we propose a principled technique for generating an optimal program obfuscation scheme by adopting a reinforcement learning approach. Given a program and a set of obfuscation transformations, a reinforcement learning model is progressively trained to select a sequence of obfuscation transformations, such that applying each transformation in order towards the program yields the optimal obfuscation result, making programs dissimilar while retaining reasonable instrumentation overhead. Our implementation can directly work on raw binary executables without source code, and our evaluation demonstrates that the trained models can effectively obfuscate executable files with low cost.

Crowdsensing has found a variety of applications (e.g., spectrum sensing, environmental monitoring) by leveraging the "wisdom" of a potentially large crowd of mobile users as "workers". The value of data collected in crowdsensing heavily depends on the quality of data provided by the workers participating in a crowdsensing task. In general, the quality of data varies for different workers. To fully exploit the potential of crowdsensing, it is important for the crowdsensing requester to know workers’ data quality, based on which the requester allocates tasks to workers and aggregates data from workers. Such quality-aware crowdsensing can greatly improve the value and usefulness of data in crowdsensing. However, the quality of workers’ data is often unknown to the requester (due to, e.g., workers’ characteristics are unknown). In this paper, under a dynamic multi-task crowdsensing framework, we devise an online data quality learning algorithm that learns the data quality of workers from their data on the fly, while making use of the learned quality information to perform task allocation and data aggregation. Compared to prior online learning algorithms (such as those for the multi-armed bandit problems), our algorithm needs to overcome the challenge that the ground truth of the interested variable is unknown. We show that under some mild conditions, our algorithm converges to the offline optimal strategy over time, and have a regret in the order of O(logt) compared to the offline optimal strategy. We provide bounds on the regret for both the requester’s utility and cost, and for both the simple average rule and the weighted average rule. We demonstrate the efficiency of the algorithm using simulation results.

The English to Chinese automatic translators cannot always satisfy the requirement of users especially in technical fields. Through research, the main reason is wrong translation of technical terminologies. The researchers summarized common mistakes of technical translation. Then developed an IE (Industrial Electronics) dictionary that is a professional technical dictionary correcting the wrong translations by Google translator. With the Python platform, researchers can automatically replaces incorrect translation by Google translator to get improved translations. Finally, we develop Web pages to show translation results that can help Chinese engineers read industrial electronic papers and this approach can be also used in other technical areas.