Juan Carlos Arceo

Working on control, motor tuning, and performance optimization of autonomous mobile robots (AMRs). * Increased the payload capacity of the E24W platform by 30% (1.2 t → 1.5 t) through control-level motor tuning and system validation. * Developing a real-time payload parameter estimation algorithm to estimate mass, friction, and inertia, enabling adaptive control under varying load conditions. * Tuned motor controllers and acceleration/torque limiters to safely extend operational limits without hardware changes. * Validated performance improvements through physical testing and safety-critical scenarios.

Senior Robotics Software Engineer at a fast-moving Danish UAV startup operating from a rural location, where my role evolved from module-level development into broader system integration and delivery responsibilities. I worked across control, safety, calibration, and operator tooling to support a UAV-based wind-turbine repair platform under frequent external demo timelines. Key responsibilities: * Held responsibility for the design, implementation, and integration of the Process Controller and several robotic subsystems (BNS, BAS, SRM, SGM) as part of the system delivered for external demos in Nov 2024, Jan 2025, and Jul 2025. * Led technical integration efforts across motor control (Dynamixel), sensor calibration (Hall sensors, optical flow), pump integrated with a UR5, and filling-station connectivity. * Developed operator-facing graphical tools to support calibration, testing, diagnostics, and routine system operation. * Investigated and resolved complex issues spanning hardware, firmware, and software layers, including root-cause analysis of video pipeline failures involving firmware-level behavior. * Supported system continuity during team changes through documentation, code stabilization, and structured handover of critical components. * Contributed to balancing HIL/SIL validation activities, safety considerations, and field readiness within a rapidly evolving development environment. Outcome: delivery of stable demo-capable systems under tight timelines, reduced integration risk, improved calibration workflows, and tooling that enabled reliable operation by non-developer users.

I guide aspiring data scientists, offering insights on statistical analysis, machine learning, and data visualization. I support mentees in their academic and professional development.

I am developing a model to forecast energy prices using diverse datasets. My role involves data collection, analysis, and the application of machine learning and data processing techniques, such as principal component analysis, to enhance prediction accuracy and support strategic decision-making.

Designed and implemented real-time control schemes for a lower-limb exoskeleton using multimodal sensors. Designed and tested a robust sensor fusion algorithm to estimate the hip and knee angular position with force-sensing resistors and inertial measurement units, the algorithm is based on a combination of artificial neural networks and signal processing. Developed graphic user interfaces in Python using Pyside, Qt5, ROS2 and Bluetooth libraries to display and record real-time experimental data for controlling and interacting with different embedded devices (Arduino/Raspberry Pi). Gave some lectures on Dynamics to bachelor students at the university (book: Meriam, J. L., Kraige, L. G., & Bolton, J. N. (2020). Engineering mechanics: dynamics. John Wiley & Sons.).

Developed a new mathematical model for analyzing mutualistic interactions between two bacteria species within a chemostat. Performed a detailed analysis of the model to establish the conditions under which the system reaches equilibrium, that is, local and global stability. Optimizing bacterial productivity within the chemostat, proposed methods to maximize bacterial growth by adjusting the dilution rates periodically and using an interval detector to estimate the optimal productivity levels. Formulated an interval detector to estimate the upper and lower bounds of unmeasured states. Designed an observer using non-monotone Lyapunov functions for enhanced monitoring and estimation of biomass concentration in heterotrophic microalgae cultures.

Created a BCI that combines electromyography (EMG) and electroencephalography (EEG) signals with biomechanical feedback to control a motorized rehabilitation device for the ankle, named motoBOTTE. Formulated several mathematical models of the motoBOTTE device, which is a one-degree-of-freedom parallel robot designed for ankle rehabilitation. Developed and Implemented several control strategies, including computed torque and active disturbance rejection control (ADRC) for parallel robots represented by differential-algebraic equations. The control strategies and models are validated through real-time implementation, demonstrating the practical applicability of the theoretical frameworks developed in the thesis.

Introduced innovative techniques for analyzing nonlinear singular systems using convex optimization, addressing challenges associated with their non-standard behaviour and stability. Designed novel control schemes to stabilize and optimize the performance of systems described by nonlinear singular differential equations, employing convex optimization methods to handle the inherent nonlinearities. Simulation studies of singular systems, implementing various theoretical concepts developed throughout the research to demonstrate their practical applicability and effectiveness in stability analysis.

Explored different control techniques for the Twin Rotor MIMO System using nonlinear control approaches and Linear Matrix Inequalities (LMIs). The LMIs were used to design robust controllers and state observers. These methodologies were applied to the highly nonlinear dynamics of the Twin Rotor system, showcasing how modern control techniques can be used to manage complex multivariable systems effectively. The Takagi-Sugeno fuzzy model developed accurately captures the nonlinear behaviour of the Twin Rotor, allowing precise and adaptable control strategies. By integrating state observer designs that utilize LMIs, the work addressed the challenge of controlling systems with partially unobservable states. The real-time application and validation of the control strategies demonstrated their effectiveness, underscoring the enhanced performance and stability of advanced electromechanical systems in dynamic environments.

The work was focused on developing a hybrid brain-computer interface (BCI) integrating electrophysiological signals for personalized ankle rehabilitation using the motoBOTTE, a motorized ergometer. The study was split into two main themes: first, the modelling and control of the rehabilitation device for tracking predefined trajectories; second, integrating human interaction with the system, utilizing models for monitoring human ankle motion and creating dynamic trajectories for the assistive robot. The thesis explored different mathematical models of the motoBOTTE device, real-time controllers to follow trajectories and human-robot interaction. A novel approach using EMG signals for real-time estimation of ankle force and using EEG signals for initiating robot movements based on motor imagery was also presented. These advancements in assistive rehabilitation technology demonstrate the potential improvements in therapy outcomes for patients with neurological impairments.

New methods were developed leveraging convex optimization techniques to address the complexities of nonlinear singular systems. The focus was on refining mathematical tools and frameworks essential for managing the unique challenges these systems present, including their non-standard behaviours and stability issues. Linear Matrix Inequalities (LMIs) were extensively utilized to design robust controllers that ensure system stability and enhance performance. This approach allowed for the extension of traditional control methodologies by solving optimization problems tailored to the needs of singular systems. The theoretical advancements achieved are complemented by practical applications, demonstrating the effectiveness of these methods in real-world scenarios where precise control of complex nonlinear dynamics is crucial. The work bridges the gap between complex theoretical concepts and their practical implementation, providing robust solutions for advanced engineering applications.

Explored control strategies for the Twin Rotor MIMO System, emphasizing the application of Linear Matrix Inequalities (LMIs). The research utilized the Takagi-Sugeno fuzzy model to handle the system's nonlinear dynamics and integrate state observer designs to manage partially unobservable states effectively. Implementing these strategies in real-time highlights the practical advantages of nonlinear approaches, offering robust solutions for complex multivariable systems and enhancing both performance and stability in dynamic environments.