Andreas Schwung

South Westphalia University of Applied Sciences

Unclaimed · AI-structured profileORCID OpenAlex

Data-driven control researcher spanning nonlinear feedback linearization, manufacturing RL, and multi-drone MPC

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

No publication-volume data is available in this snapshot, so output is unscored. This axis would measure total peer-reviewed and preprint output over time, productivity trends, and venue distribution — none of which can be inferred from three cataloged papers.

Impact—

Impact is unscored: citation counts, h-index, and adoption metrics are absent from the input data. This axis would measure how often the work is cited and built upon — whether others pick up the variable-horizon DMPC or inverse-model MBRL methods.

Focus0/100

The focus score of 0 reflects a perfectly even spread: the raw field distribution assigns exactly one third each to multi-agent motion planning, reinforcement learning for manufacturing control, and nonlinear control with machine learning. No single field dominates. This indicates a deliberately breadth-oriented agenda across three adjacent control-and-learning domains rather than concentration in one niche — breadth visible across all three cataloged 2026 papers.

Collaboration41/100

Collaboration scores 41, consistent with a compact network: the cataloged papers involve seven distinct co-authors, with recurring partners such as Linda Mümken, Michael Schwung, Steve Yuwono, and Dorothea Schwung. Authorship roles alternate between first author on the manufacturing RL paper and last author on the two control papers, indicating both hands-on contribution and senior supervisory positions within a small, stable group.

Activity—

Activity is unscored: recent publication cadence and submission continuity are not in the available data. The axis would track research tempo — whether the 2026 cluster of three arXiv papers reflects a sustained pace or a single burst.

Reproducibility55/100

Reproducibility scores 55, the mean across three rated papers, each graded medium readiness. Evolve's verdicts converge on one pattern: technically careful work with stated stability or feasibility guarantees, but evidence limited to single simulations, no baseline comparisons, single-seed runs, and simulation-only swaps on one plant. Releasing code, baselines, and hardware results would most directly raise this axis.

Research summary

Andreas Schwung, based at South Westphalia University of Applied Sciences, works at the intersection of nonlinear control theory and machine learning, with this profile name-resolved from public bibliographic data (ORCID 0000-0001-8405-0977). The Evolve-reviewed snapshot spans three 2026 arXiv preprints spread evenly across three fields — multi-agent motion planning, reinforcement learning for manufacturing control, and nonlinear control with machine learning — each accounting for one third of the cataloged output. His authorship pattern mixes hands-on execution (first author on the distributed model-based reinforcement learning paper for modular production systems) with senior supervision (last author on the multi-drone distributed MPC and neural feedback linearization papers). Technically, his work pursues one consistent theme: embedding structure and stability guarantees into learned controllers, whether through augmented-Lagrangian relative-degree constraints with practical stability bounds, frozen inverse models inside MBRL policies, or conflict-predictive variable horizons preserving recursive feasibility and asymptotic stability. Evolve rates all three papers medium on reproducibility readiness (mean 55 across three rated papers), with recurring limitations of single simulations, no baselines, single-seed runs, and simulation-only evidence. Collaboration remains compact — seven distinct co-authors in a recurring group — suggesting a stable, locally anchored team. Citation-level impact and productivity metrics are not yet available in this snapshot, so impact and output axes remain unscored.

Collaboration pattern

Collaboration is anchored in a compact, recurring group: seven distinct co-authors appear across the cataloged work, including Linda Mümken, Michael Schwung, Stefan Lier, Steve Yuwono, Sofiene Lassoued, Dorothea Schwung, and Lakshmi Priya P. K. The pattern suggests a stable, institutionally rooted team at South Westphalia University of Applied Sciences, with Schwung alternating between first-author execution on the manufacturing RL paper and last-author seniority on the two control theory papers.

Research keywords

Data-driven controlFeedback linearizationNeural Lie derivativesPractical stabilityModel-based reinforcement learningInverse modelsModular manufacturingDistributed optimizationModel predictive controlMulti-drone collision avoidanceRecursive feasibilityMulti-agent systems

Highlights

Neural feedback linearization with stability guarantees via augmented Lagrangian
The 2026 preprint adds augmented-Lagrangian relative-degree constraints to neural feedback linearization, yielding a practical stability bound learned from data. Evolve calls the idea promising but notes a single simulation and no baselines.
Freezing inverse models inside MBRL for modular production systems
As first author, he proposes freezing learned inverse models within model-based RL policies so agents plan in task space for distributed optimization of modular manufacturing. Simulation results on one plant are promising, though single-seed and simulation-only.
Conflict-predictive variable horizons for multi-drone DMPC
As last author, he contributes variable prediction horizons with decoupled guarantees to distributed MPC for multi-drone collision avoidance, preserving recursive feasibility and asymptotic stability. Evolve rates the theory solid but notes reliance on a prior framework and simulation-only validation.

Timeline

2026First-author preprint proposes distributed model-based reinforcement learning with inverse models for optimizing modular production systems.
2026Last-author preprint introduces conflict-predictive variable horizons for multi-drone distributed model predictive control.
2026Last-author preprint learns neural feedback linearization under augmented-Lagrangian constraints with a practical stability bound.

Papers in the Evolve catalog

Frequent co-authors

Linda Mümken×1 Michael Schwung×1 Stefan Lier×1 Steve Yuwono×1 Sofiene Lassoued×1 Dorothea Schwung×1 Lakshmi Priya P. K.×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric signals derive solely from Evolve's reviewed snapshot (3 arXiv preprints, all 2026); citation counts, h-index, and total output are unavailable, so output, impact, and activity axes are unscored.
All cataloged evidence is simulation-only; no peer-reviewed venue, hardware validation, or experimental replication data is in scope.
Co-author statistics (7 distinct co-authors) reflect only the cataloged papers and may understate the full collaboration network.
Country-level affiliation metadata is missing in the source record; identity is resolved via ORCID 0000-0001-8405-0977 from public bibliographic data.