Distributed Covariance Steering via Non-Convex ADMM for Large-Scale Multi-Agent Systems

Arshiya Taj Abdul; Augustinos D. Saravanos; Efstathios Bakolas; Evangelos A. Theodorou; Isin M. Balci

arxiv: 2604.04499 · v1 · submitted 2026-04-06 · 📡 eess.SY · cs.SY· math.OC

Distributed Covariance Steering via Non-Convex ADMM for Large-Scale Multi-Agent Systems

Augustinos D. Saravanos , Isin M. Balci , Arshiya Taj Abdul , Efstathios Bakolas , Evangelos A. Theodorou This is my paper

Pith reviewed 2026-05-10 20:05 UTC · model grok-4.3

classification 📡 eess.SY cs.SYmath.OC

keywords distributed covariance steeringADMMmulti-agent systemsprobabilistic collision avoidancestochastic controlconsensus optimizationnon-convex optimizationGaussian distributions

0 comments

The pith

Distributed ADMM variants solve covariance steering for thousands of multi-agent systems under probabilistic collision constraints with new convergence proofs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops three distributed ADMM-based methods for steering the means and covariances of large numbers of stochastic linear agents between Gaussian distributions while satisfying probabilistic collision avoidance. Full-covariance consensus yields the safest but most expensive solutions, partial-covariance consensus exchanges less data, and mean-only consensus maximizes scalability. The central advance is a convergence analysis showing that the partial and mean variants reach stationary points despite the non-convex constraints being linearized at each iteration, with full consensus inheriting standard ADMM guarantees. This matters because centralized planning becomes intractable at scale, while these methods let agents coordinate safely with limited communication.

Core claim

The paper establishes novel convergence guarantees for distributed ADMM applied to consensus problems with iteratively linearized non-convex constraints. These guarantees prove that the partial-covariance and mean-consensus distributed covariance steering methods converge to stationary points, while full-covariance consensus follows from existing ADMM theory. The resulting algorithms steer large-scale multi-agent stochastic linear systems with Gaussian distributions between prescribed terminal distributions while enforcing probabilistic safety.

What carries the argument

The family of Distributed Covariance Steering (DCS) algorithms based on ADMM, consisting of Full-Covariance-Consensus DCS, Partial-Covariance-Consensus DCS, and Mean-Consensus DCS, each paired with successive linearization of the probabilistic collision constraints.

If this is right

The partial and mean consensus variants remain provably convergent even though the safety constraints are non-convex.
Agents need only exchange partial covariance information or means to maintain safety, lowering communication and computation.
The methods scale to thousands of agents in two- and three-dimensional simulations while keeping trajectories safe.
Full consensus yields the least conservative solutions but at higher cost, allowing users to choose the appropriate trade-off.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The partial-consensus idea could be reused for other multi-agent planning problems where only certain statistics need to be coordinated.
The convergence result may apply directly to broader classes of consensus optimization with linearized non-convex constraints beyond covariance steering.
Real-robot experiments would test whether the Gaussian and linearity assumptions survive sensor noise and model mismatch.

Load-bearing premise

Iteratively linearizing the non-convex probabilistic collision constraints still produces stationary points that satisfy the original safety requirements for stochastic linear systems with Gaussian uncertainty.

What would settle it

A concrete counter-example system in which a converged stationary point of any of the three DCS methods violates the original probabilistic collision avoidance probability bounds.

read the original abstract

This paper studies the problem of steering large-scale multi-agent stochastic linear systems between Gaussian distributions under probabilistic collision avoidance constraints. We introduce a family of \textit{distributed covariance steering (DCS)} methods based on the Alternating Direction Method of Multipliers (ADMM), each offering different trade-offs between conservatism and computational efficiency. The first method, Full-Covariance-Consensus (FCC)-DCS, enforces consensus over both the means and covariances of neighboring agents, yielding the least conservative safe solutions. The second approach, Partial-Covariance-Consensus (PCC)-DCS, leverages the insight that safety can be maintained by exchanging only partial covariance information, reducing computational demands. The third method, Mean-Consensus (MC)-DCS, provides the most scalable alternative by requiring consensus only on mean states. Furthermore, we establish novel convergence guarantees for distributed ADMM with iteratively linearized non-convex constraints, covering a broad class of consensus optimization problems. This analysis proves convergence to stationary points for PCC-DCS and MC-DCS, while the convergence of FCC-DCS follows from standard ADMM theory. Simulations in 2D and 3D multi-agent environments verify safety, illustrate the trade-offs between methods, and demonstrate scalability to thousands of agents.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Practical distributed ADMM methods for multi-agent covariance steering with varying consensus levels and convergence analysis for linearized non-convex cases, but safety for the original constraints rests mostly on simulations rather than proof.

read the letter

The paper presents three distributed covariance steering approaches using ADMM for large-scale multi-agent stochastic linear systems with probabilistic collision avoidance. The punchline is that they offer different consensus levels—full covariance, partial covariance, and mean-only—to balance safety conservatism against computation and communication costs, along with convergence guarantees for the non-convex linearized versions. They do a solid job developing the methods. Full-covariance consensus gives the safest but most expensive solutions. Partial consensus exchanges only some covariance info, which cuts down on overhead while keeping safety. Mean consensus is the lightest but most conservative. The simulations in 2D and 3D environments confirm that all three maintain safety and scale to thousands of agents, which shows the practical value for robotics and autonomous systems. The convergence analysis is new for the partial and mean cases. They use iterative linearization of the non-convex constraints and prove convergence to stationary points under ADMM. This covers a broader class of consensus problems than just this application. The main soft spot is whether those stationary points actually satisfy the original non-convex safety constraints. The linearization happens around the current iterate each time, so at convergence you have first-order stationarity for the last linear approximation. But the true constraints involve nonlinear functions of the covariances, like overlap probabilities for Gaussians. There's no argument in the work that the error vanishes or that the limit point meets the nonlinear inequalities. Safety is checked in the simulations, but that doesn't replace a theoretical guarantee. If the linearization doesn't stay accurate, the agents might not be as safe as claimed in theory. The work assumes linear dynamics and Gaussian distributions, which is common but restricts it to that setting. The citation pattern seems standard for ADMM and covariance steering literature. This is aimed at researchers in stochastic control, optimization, and multi-agent systems who need scalable distributed methods. It has enough novelty in the variants and analysis, plus empirical results, to go to peer review. Reviewers can dig into the linearization details and see if the safety holds up beyond simulations. I recommend sending it for review.

Referee Report

1 major / 1 minor

Summary. The paper proposes three distributed ADMM-based covariance steering methods (FCC-DCS, PCC-DCS, MC-DCS) for large-scale multi-agent stochastic linear systems subject to probabilistic collision avoidance constraints. FCC-DCS enforces full covariance consensus, PCC-DCS uses partial covariance information, and MC-DCS requires only mean consensus. Novel convergence guarantees are claimed for distributed ADMM with iteratively linearized non-convex constraints, proving convergence to stationary points for PCC-DCS and MC-DCS (with FCC-DCS following from standard ADMM theory). Simulations in 2D/3D environments verify safety and demonstrate scalability to thousands of agents.

Significance. If the central theoretical claims hold, the work provides valuable trade-offs between conservatism and scalability for safe multi-agent control, along with new convergence results for non-convex distributed optimization. The simulation results on large-scale systems and explicit handling of Gaussian distributions are practical strengths. The iterative linearization approach for non-convex constraints could extend to other consensus problems if the safety bridge is rigorously established.

major comments (1)

[Abstract and theoretical analysis] Abstract (convergence guarantees paragraph): The analysis establishes convergence to stationary points of the iteratively linearized problems for PCC-DCS and MC-DCS. However, no argument is given that these points satisfy the original nonlinear probabilistic collision avoidance constraints (typically involving log-determinant or quadratic forms on covariances). Since each linearization occurs around the current iterate, the limit satisfies first-order stationarity only for the final surrogate; without vanishing linearization error, constraint qualification, or exact penalty analysis, the safety claims rest on an unproven connection between the approximated and true problems.

minor comments (1)

The abstract and introduction could more explicitly state the form of the probabilistic collision avoidance constraints and the linearization procedure to aid readability.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful and constructive review of our manuscript. We address the major comment point-by-point below and will revise the paper to strengthen the presentation of the theoretical results.

read point-by-point responses

Referee: [Abstract and theoretical analysis] Abstract (convergence guarantees paragraph): The analysis establishes convergence to stationary points of the iteratively linearized problems for PCC-DCS and MC-DCS. However, no argument is given that these points satisfy the original nonlinear probabilistic collision avoidance constraints (typically involving log-determinant or quadratic forms on covariances). Since each linearization occurs around the current iterate, the limit satisfies first-order stationarity only for the final surrogate; without vanishing linearization error, constraint qualification, or exact penalty analysis, the safety claims rest on an unproven connection between the approximated and true problems.

Authors: We thank the referee for highlighting this important distinction. Our analysis proves that the distributed ADMM sequence converges to a first-order stationary point of the sequence of successively linearized problems for PCC-DCS and MC-DCS (with FCC-DCS following from standard ADMM theory). We agree that an explicit argument linking this limit to satisfaction of the original nonlinear probabilistic constraints is not fully developed in the current manuscript. In the revision we will add a clarifying remark in the theoretical section noting that, at convergence, the linearization is performed at the limit point itself, so the stationarity condition holds for the original constraint gradient when the linearization error vanishes. We will also emphasize that the safety guarantees are supported by the extensive numerical experiments, which verify that the returned trajectories satisfy the original chance constraints to the prescribed probability level across all three methods. revision: yes

Circularity Check

0 steps flagged

No significant circularity in claimed convergence analysis or method derivations

full rationale

The paper's core claims rest on introducing three DCS variants (FCC, PCC, MC) and providing new convergence guarantees for distributed ADMM applied to iteratively linearized non-convex probabilistic constraints. These guarantees are presented as novel analysis covering a broad class of consensus problems, with FCC-DCS falling back to standard ADMM theory. No equations or steps in the abstract or described contributions reduce a 'prediction' or stationary-point result to a fitted input, self-definition, or prior self-citation by construction. The linearization is an explicit algorithmic choice whose convergence properties are analyzed separately rather than assumed equivalent to the original nonlinear problem. The derivation chain therefore remains self-contained against external benchmarks and does not exhibit the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no explicit free parameters, axioms, or invented entities; the methods rely on standard stochastic linear dynamics and Gaussian assumptions common to the field.

pith-pipeline@v0.9.0 · 5551 in / 1126 out tokens · 76123 ms · 2026-05-10T20:05:32.245975+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

54 extracted references · 54 canonical work pages

[1]

Optimal time trajectory and coordination for connected and automated vehicles,

A. A. Malikopoulos, L. Beaver, and I. V . Chremos, “Optimal time trajectory and coordination for connected and automated vehicles,” Automatica, vol. 125, p. 109469, 2021. Fig. 7:Large-scale 2D task with 1024 agents via MC-DCS. Snapshot at time instantk= 20. 101 102 103 N 101 102 103 Computational Time (s) Centralized CS FCC-DCS PCC-DCS MC-DCS Fig. 8:Scala...

work page 2021
[2]

Prediction, planning, and coordination of thousand-warehousing-robot networks with motion and communication uncertainties,

Z. Liu, H. Wang, H. Wei, M. Liu, and Y .-H. Liu, “Prediction, planning, and coordination of thousand-warehousing-robot networks with motion and communication uncertainties,”IEEE Transactions on Automation Science and Engineering, vol. 18, no. 4, pp. 1705–1717, 2020

work page 2020
[3]

Distributed formation control of drones with onboard perception,

K. M. Kabore and S. G ¨uler, “Distributed formation control of drones with onboard perception,”IEEE/ASME Transactions on Mechatronics, pp. 1–11, 2021

work page 2021
[4]

Coordinated control of multi-robot sys- tems: A survey,

J. Cort ´es and M. Egerstedt, “Coordinated control of multi-robot sys- tems: A survey,”SICE Journal of Control, Measurement, and System Integration, vol. 10, no. 6, pp. 495–503, 2017

work page 2017
[5]

The scenario approach for stochastic model predictive control with bounds on closed- 16 loop constraint violations,

G. Schildbach, L. Fagiano, C. Frei, and M. Morari, “The scenario approach for stochastic model predictive control with bounds on closed- 16 loop constraint violations,”Automatica, vol. 50, no. 12, pp. 3009–3018, 2014

work page 2014
[6]

Stochastic MPC with robustness to bounded parameteric uncertainty,

E. Arcari, A. Iannelli, A. Carron, and M. N. Zeilinger, “Stochastic MPC with robustness to bounded parameteric uncertainty,”IEEE Transactions on Automatic Control, vol. 68, no. 12, pp. 7601–7615, 2023

work page 2023
[7]

Stochastic linear model predictive control with chance constraints–a review,

M. Farina, L. Giulioni, and R. Scattolini, “Stochastic linear model predictive control with chance constraints–a review,”Journal of Process Control, vol. 44, pp. 53–67, 2016

work page 2016
[8]

Optimal transport in systems and control,

Y . Chen, T. T. Georgiou, and M. Pavon, “Optimal transport in systems and control,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, no. 1, pp. 89–113, 2021

work page 2021
[9]

Optimal steering of a linear stochastic system to a final prob- ability distribution, Part I,

——, “Optimal steering of a linear stochastic system to a final prob- ability distribution, Part I,”IEEE Transactions on Automatic Control, vol. 61, no. 5, pp. 1158–1169, 2015

work page 2015
[10]

Finite-horizon covariance control for discrete-time stochas- tic linear systems subject to input constraints,

E. Bakolas, “Finite-horizon covariance control for discrete-time stochas- tic linear systems subject to input constraints,”Automatica, vol. 91, pp. 61–68, 2018

work page 2018
[11]

Optimal covariance steering for discrete-time linear stochastic systems,

F. Liu, G. Rapakoulias, and P. Tsiotras, “Optimal covariance steering for discrete-time linear stochastic systems,”IEEE Transactions on Automatic Control, vol. 70, no. 4, pp. 2289–2304, 2025

work page 2025
[12]

Nonlinear uncertainty con- trol with iterative covariance steering,

J. Ridderhof, K. Okamoto, and P. Tsiotras, “Nonlinear uncertainty con- trol with iterative covariance steering,” in2019 IEEE 58th Conference on Decision and Control (CDC). IEEE, 2019, pp. 3484–3490

work page 2019
[13]

Oper- ator splitting covariance steering for safe stochastic nonlinear control,

A. Ratheesh, V . Pacelli, A. D. Saravanos, and E. A. Theodorou, “Oper- ator splitting covariance steering for safe stochastic nonlinear control,” in2025 IEEE 64th Conference on Decision and Control (CDC), 2025, pp. 3552–3559

work page 2025
[14]

Convex optimization for finite-horizon robust covariance control of linear stochastic systems,

G. Kotsalis, G. Lan, and A. S. Nemirovski, “Convex optimization for finite-horizon robust covariance control of linear stochastic systems,” SIAM Journal on Control and Optimization, vol. 59, no. 1, pp. 296– 319, 2021

work page 2021
[15]

Data-driven covariance steering control design,

J. Pilipovsky and P. Tsiotras, “Data-driven covariance steering control design,” in2023 62nd IEEE Conference on Decision and Control (CDC). IEEE, 2023, pp. 2610–2615

work page 2023
[16]

Density steering of Gaussian mixture models for discrete-time linear systems,

I. M. Balci and E. Bakolas, “Density steering of Gaussian mixture models for discrete-time linear systems,” in2024 American Control Conference (ACC). IEEE, 2024, pp. 3935–3940

work page 2024
[17]

Distribution steering for discrete- time uncertain ensemble systems,

G. Wu, P. Tsiotras, and A. Lindquist, “Distribution steering for discrete- time uncertain ensemble systems,”arXiv preprint arXiv:2405.12415, 2024

work page arXiv 2024
[18]

Trajectory distribution control for model predictive path integral control using covariance steering,

J. Yin, Z. Zhang, E. Theodorou, and P. Tsiotras, “Trajectory distribution control for model predictive path integral control using covariance steering,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 1478–1484

work page 2022
[19]

Covariance steering for uncertain contact-rich systems,

Y . Shirai, D. K. Jha, and A. U. Raghunathan, “Covariance steering for uncertain contact-rich systems,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 7923–7929

work page 2023
[20]

Sequential chance-constrained covariance steering for robust cislunar trajectory design under uncertainties,

N. Kumagai and K. Oguri, “Sequential chance-constrained covariance steering for robust cislunar trajectory design under uncertainties,” in AAS/AIAA Astrodynamics Specialist Conference, 2024, pp. 1–19

work page 2024
[21]

Dis- tributed Covariance Steering with Consensus ADMM for Stochastic Multi-Agent Systems,

A. D. Saravanos, A. Tsolovikos, E. Bakolas, and E. Theodorou, “Dis- tributed Covariance Steering with Consensus ADMM for Stochastic Multi-Agent Systems,” inProceedings of Robotics: Science and Systems, Virtual, July 2021

work page 2021
[22]

Density control of interacting agent systems,

Y . Chen, “Density control of interacting agent systems,”IEEE Transac- tions on Automatic Control, vol. 69, no. 1, pp. 246–260, 2024

work page 2024
[23]

Steering large agent populations using mean-field schr ¨odinger bridges with gaussian mixture models,

G. Rapakoulias, A. R. Pedram, and P. Tsiotras, “Steering large agent populations using mean-field schr ¨odinger bridges with gaussian mixture models,”IEEE Control Systems Letters, 2025

work page 2025
[24]

Decentralized probabilistic den- sity control of autonomous swarms with safety constraints,

N. Demir, U. Eren, and B. Ac ¸ıkmes ¸e, “Decentralized probabilistic den- sity control of autonomous swarms with safety constraints,”Autonomous Robots, vol. 39, no. 4, pp. 537–554, 2015

work page 2015
[25]

Group steering: Approaches based on power moments,

G. Wu and A. Lindquist, “Group steering: Approaches based on power moments,”arXiv preprint arXiv:2211.13370, 2022

work page arXiv 2022
[26]

Distributed model predictive covariance steering,

A. D. Saravanos, I. M. Balci, E. Bakolas, and E. A. Theodorou, “Distributed model predictive covariance steering,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024, pp. 5740–5747

work page 2024
[27]

Distributed Hierarchical Dis- tribution Control for Very-Large-Scale Clustered Multi-Agent Systems,

A. D. Saravanos, Y . Li, and E. Theodorou, “Distributed Hierarchical Dis- tribution Control for Very-Large-Scale Clustered Multi-Agent Systems,” inProceedings of Robotics: Science and Systems, Daegu, Republic of Korea, July 2023

work page 2023
[28]

Robust motion coordi- nation with covariance steering model predictive control in bandwidth limited scenarios,

H. Bai, H. Zhu, X. Zhao, H. Li, and Y . Wang, “Robust motion coordi- nation with covariance steering model predictive control in bandwidth limited scenarios,” in2024 IEEE International Conference on Robotics and Biomimetics (ROBIO), 2024, pp. 1809–1814

work page 2024
[29]

Distributed optimization and statistical learning via the alternating direction method of multipliers,

S. Boyd, N. Parikh, E. Chu, B. Peleato, and J. Eckstein, “Distributed optimization and statistical learning via the alternating direction method of multipliers,”Foundations and Trends® in Machine Learning, vol. 3, no. 1, pp. 1–122, 2011

work page 2011
[30]

Distributed differential dynamic programming architectures for large-scale multia- gent control,

A. D. Saravanos, Y . Aoyama, H. Zhu, and E. A. Theodorou, “Distributed differential dynamic programming architectures for large-scale multia- gent control,”IEEE Transactions on Robotics, vol. 39, no. 6, pp. 4387– 4407, 2023

work page 2023
[31]

Distributed opti- mization methods for multi-robot systems: Part 1—a tutorial [tutorial],

O. Shorinwa, T. Halsted, J. Yu, and M. Schwager, “Distributed opti- mization methods for multi-robot systems: Part 1—a tutorial [tutorial],” IEEE Robotics & Automation Magazine, vol. 31, no. 3, pp. 121–138, 2024

work page 2024
[32]

Scalable robust optimization for safe multi-agent control under unknown deterministic uncertainty,

A. T. Abdul, A. D. Saravanos, and E. A. Theodorou, “Scalable robust optimization for safe multi-agent control under unknown deterministic uncertainty,” in2025 American Control Conference (ACC). IEEE, 2025, pp. 3666–3673

work page 2025
[33]

Global convergence of splitting methods for nonconvex composite optimization,

G. Li and T. K. Pong, “Global convergence of splitting methods for nonconvex composite optimization,”SIAM Journal on Optimization, vol. 25, no. 4, pp. 2434–2460, 2015

work page 2015
[34]

Convergence of alternating direction method for minimizing sum of two nonconvex functions with linear constraints,

K. Guo, D. Han, and T.-T. Wu, “Convergence of alternating direction method for minimizing sum of two nonconvex functions with linear constraints,”International Journal of Computer Mathematics, vol. 94, no. 8, pp. 1653–1669, 2017

work page 2017
[35]

Convergence analysis of alternating direction method of multipliers for a family of nonconvex problems,

M. Hong, Z.-Q. Luo, and M. Razaviyayn, “Convergence analysis of alternating direction method of multipliers for a family of nonconvex problems,”SIAM Journal on Optimization, vol. 26, no. 1, pp. 337–364, 2016

work page 2016
[36]

Linearized ADMM for nonconvex non- smooth optimization with convergence analysis,

Q. Liu, X. Shen, and Y . Gu, “Linearized ADMM for nonconvex non- smooth optimization with convergence analysis,”IEEE access, vol. 7, pp. 76 131–76 144, 2019

work page 2019
[37]

Linearized ADMM converges to second-order stationary points for non-convex problems,

S. Lu, J. D. Lee, M. Razaviyayn, and M. Hong, “Linearized ADMM converges to second-order stationary points for non-convex problems,” IEEE Transactions on Signal Processing, vol. 69, pp. 4859–4874, 2021

work page 2021
[38]

Iteration-complexity of a linearized proximal multiblock ADMM class for linearly constrained nonconvex optimization problems,

J. G. Melo and R. Monteiro, “Iteration-complexity of a linearized proximal multiblock ADMM class for linearly constrained nonconvex optimization problems,”Available on: http://www. optimization-online. org, 2017

work page 2017
[39]

Global convergence of ADMM in nonconvex nonsmooth optimization,

Y . Wang, W. Yin, and J. Zeng, “Global convergence of ADMM in nonconvex nonsmooth optimization,”Journal of Scientific Computing, vol. 78, pp. 29–63, 2019

work page 2019
[40]

Douglas–rachford splitting and admm for nonconvex optimization: Tight convergence results,

A. Themelis and P. Patrinos, “Douglas–rachford splitting and admm for nonconvex optimization: Tight convergence results,”SIAM Journal on Optimization, vol. 30, no. 1, pp. 149–181, 2020

work page 2020
[41]

A two-level distributed algorithm for nonconvex constrained optimization,

K. Sun and X. A. Sun, “A two-level distributed algorithm for nonconvex constrained optimization,”Computational Optimization and Applica- tions, vol. 84, no. 2, pp. 609–649, 2023

work page 2023
[42]

A two-level ADMM algorithm for AC OPF with global conver- gence guarantees,

——, “A two-level ADMM algorithm for AC OPF with global conver- gence guarantees,”IEEE Transactions on Power Systems, vol. 36, no. 6, pp. 5271–5281, 2021

work page 2021
[43]

Fast and stable nonconvex constrained distributed optimization: the ELLADA algorithm,

W. Tang and P. Daoutidis, “Fast and stable nonconvex constrained distributed optimization: the ELLADA algorithm,”Optimization and Engineering, vol. 23, no. 1, pp. 259–301, 2022

work page 2022
[44]

Constrained minimum variance and covariance steering based on affine disturbance feedback control pa- rameterization,

I. M. Balci and E. Bakolas, “Constrained minimum variance and covariance steering based on affine disturbance feedback control pa- rameterization,”International Journal of Robust and Nonlinear Control, vol. 34, no. 11, pp. 7332–7370, 2024

work page 2024
[45]

Optimal stochastic vehicle path planning using covariance steering,

K. Okamoto and P. Tsiotras, “Optimal stochastic vehicle path planning using covariance steering,”IEEE Robotics and Automation Letters, vol. 4, no. 3, pp. 2276–2281, 2019

work page 2019
[46]

Discrete-time optimal covariance steering via semidefinite programming,

G. Rapakoulias and P. Tsiotras, “Discrete-time optimal covariance steering via semidefinite programming,” in2023 62nd IEEE Conference on Decision and Control (CDC), 2023, pp. 1802–1807

work page 2023
[47]

Nocedal and S

J. Nocedal and S. J. Wright,Numerical optimization. Springer, 2006

work page 2006
[48]

Sequential quadratic programming,

P. T. Boggs and J. W. Tolle, “Sequential quadratic programming,”Acta numerica, vol. 4, pp. 1–51, 1995

work page 1995
[49]

On the global and linear convergence of the generalized alternating direction method of multipliers,

W. Deng and W. Yin, “On the global and linear convergence of the generalized alternating direction method of multipliers,”Journal of Scientific Computing, vol. 66, pp. 889–916, 2016

work page 2016
[50]

ApS,The MOSEK optimization toolbox for MATLAB manual

M. ApS,The MOSEK optimization toolbox for MATLAB manual. Version 9.0., 2019

work page 2019
[51]

OSQP: An operator splitting solver for quadratic programs,

B. Stellato, G. Banjac, P. Goulart, A. Bemporad, and S. Boyd, “OSQP: An operator splitting solver for quadratic programs,”Mathematical Programming Computation, vol. 12, no. 4, pp. 637–672, 2020

work page 2020
[52]

Gcbf+: A neural graph control barrier function framework for distributed safe multi-agent control,

S. Zhang, O. So, K. Garg, and C. Fan, “Gcbf+: A neural graph control barrier function framework for distributed safe multi-agent control,” IEEE Transactions on Robotics, 2025. SARAVANOSet al.: DISTRIBUTED COVARIANCE STEERING VIA NON-CONVEX ADMM FOR LARGE-SCALE MULTI-AGENT SYSTEMS 17

work page 2025
[53]

Deep distributed optimization for large-scale quadratic programming,

A. D. Saravanos, H. Kuperman, A. Oshin, A. T. Abdul, V . Pacelli, and E. Theodorou, “Deep distributed optimization for large-scale quadratic programming,” inThe Thirteenth International Conference on Learning Representations, 2025. Augustinos D. Saravanos(Graduate Student Member, IEEE) received his Diploma in Elec- trical and Computer Engineering with hig...

work page 2025
[54]

In 2011, he graduated with his PhD in Computer Science from USC

He also holds three MSc degrees in Production Engineering from TUC in 2003, Computer Science and Engineering from University of Minnesota in 2007, and Electrical Engineering from the University of Southern California (USC) in 2010. In 2011, he graduated with his PhD in Computer Science from USC. From 2011 to 2013, he was a Postdoc- toral Research Fellow w...

work page 2003

[1] [1]

Optimal time trajectory and coordination for connected and automated vehicles,

A. A. Malikopoulos, L. Beaver, and I. V . Chremos, “Optimal time trajectory and coordination for connected and automated vehicles,” Automatica, vol. 125, p. 109469, 2021. Fig. 7:Large-scale 2D task with 1024 agents via MC-DCS. Snapshot at time instantk= 20. 101 102 103 N 101 102 103 Computational Time (s) Centralized CS FCC-DCS PCC-DCS MC-DCS Fig. 8:Scala...

work page 2021

[2] [2]

Prediction, planning, and coordination of thousand-warehousing-robot networks with motion and communication uncertainties,

Z. Liu, H. Wang, H. Wei, M. Liu, and Y .-H. Liu, “Prediction, planning, and coordination of thousand-warehousing-robot networks with motion and communication uncertainties,”IEEE Transactions on Automation Science and Engineering, vol. 18, no. 4, pp. 1705–1717, 2020

work page 2020

[3] [3]

Distributed formation control of drones with onboard perception,

K. M. Kabore and S. G ¨uler, “Distributed formation control of drones with onboard perception,”IEEE/ASME Transactions on Mechatronics, pp. 1–11, 2021

work page 2021

[4] [4]

Coordinated control of multi-robot sys- tems: A survey,

J. Cort ´es and M. Egerstedt, “Coordinated control of multi-robot sys- tems: A survey,”SICE Journal of Control, Measurement, and System Integration, vol. 10, no. 6, pp. 495–503, 2017

work page 2017

[5] [5]

The scenario approach for stochastic model predictive control with bounds on closed- 16 loop constraint violations,

G. Schildbach, L. Fagiano, C. Frei, and M. Morari, “The scenario approach for stochastic model predictive control with bounds on closed- 16 loop constraint violations,”Automatica, vol. 50, no. 12, pp. 3009–3018, 2014

work page 2014

[6] [6]

Stochastic MPC with robustness to bounded parameteric uncertainty,

E. Arcari, A. Iannelli, A. Carron, and M. N. Zeilinger, “Stochastic MPC with robustness to bounded parameteric uncertainty,”IEEE Transactions on Automatic Control, vol. 68, no. 12, pp. 7601–7615, 2023

work page 2023

[7] [7]

Stochastic linear model predictive control with chance constraints–a review,

M. Farina, L. Giulioni, and R. Scattolini, “Stochastic linear model predictive control with chance constraints–a review,”Journal of Process Control, vol. 44, pp. 53–67, 2016

work page 2016

[8] [8]

Optimal transport in systems and control,

Y . Chen, T. T. Georgiou, and M. Pavon, “Optimal transport in systems and control,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, no. 1, pp. 89–113, 2021

work page 2021

[9] [9]

Optimal steering of a linear stochastic system to a final prob- ability distribution, Part I,

——, “Optimal steering of a linear stochastic system to a final prob- ability distribution, Part I,”IEEE Transactions on Automatic Control, vol. 61, no. 5, pp. 1158–1169, 2015

work page 2015

[10] [10]

Finite-horizon covariance control for discrete-time stochas- tic linear systems subject to input constraints,

E. Bakolas, “Finite-horizon covariance control for discrete-time stochas- tic linear systems subject to input constraints,”Automatica, vol. 91, pp. 61–68, 2018

work page 2018

[11] [11]

Optimal covariance steering for discrete-time linear stochastic systems,

F. Liu, G. Rapakoulias, and P. Tsiotras, “Optimal covariance steering for discrete-time linear stochastic systems,”IEEE Transactions on Automatic Control, vol. 70, no. 4, pp. 2289–2304, 2025

work page 2025

[12] [12]

Nonlinear uncertainty con- trol with iterative covariance steering,

J. Ridderhof, K. Okamoto, and P. Tsiotras, “Nonlinear uncertainty con- trol with iterative covariance steering,” in2019 IEEE 58th Conference on Decision and Control (CDC). IEEE, 2019, pp. 3484–3490

work page 2019

[13] [13]

Oper- ator splitting covariance steering for safe stochastic nonlinear control,

A. Ratheesh, V . Pacelli, A. D. Saravanos, and E. A. Theodorou, “Oper- ator splitting covariance steering for safe stochastic nonlinear control,” in2025 IEEE 64th Conference on Decision and Control (CDC), 2025, pp. 3552–3559

work page 2025

[14] [14]

Convex optimization for finite-horizon robust covariance control of linear stochastic systems,

G. Kotsalis, G. Lan, and A. S. Nemirovski, “Convex optimization for finite-horizon robust covariance control of linear stochastic systems,” SIAM Journal on Control and Optimization, vol. 59, no. 1, pp. 296– 319, 2021

work page 2021

[15] [15]

Data-driven covariance steering control design,

J. Pilipovsky and P. Tsiotras, “Data-driven covariance steering control design,” in2023 62nd IEEE Conference on Decision and Control (CDC). IEEE, 2023, pp. 2610–2615

work page 2023

[16] [16]

Density steering of Gaussian mixture models for discrete-time linear systems,

I. M. Balci and E. Bakolas, “Density steering of Gaussian mixture models for discrete-time linear systems,” in2024 American Control Conference (ACC). IEEE, 2024, pp. 3935–3940

work page 2024

[17] [17]

Distribution steering for discrete- time uncertain ensemble systems,

G. Wu, P. Tsiotras, and A. Lindquist, “Distribution steering for discrete- time uncertain ensemble systems,”arXiv preprint arXiv:2405.12415, 2024

work page arXiv 2024

[18] [18]

Trajectory distribution control for model predictive path integral control using covariance steering,

J. Yin, Z. Zhang, E. Theodorou, and P. Tsiotras, “Trajectory distribution control for model predictive path integral control using covariance steering,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 1478–1484

work page 2022

[19] [19]

Covariance steering for uncertain contact-rich systems,

Y . Shirai, D. K. Jha, and A. U. Raghunathan, “Covariance steering for uncertain contact-rich systems,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 7923–7929

work page 2023

[20] [20]

Sequential chance-constrained covariance steering for robust cislunar trajectory design under uncertainties,

N. Kumagai and K. Oguri, “Sequential chance-constrained covariance steering for robust cislunar trajectory design under uncertainties,” in AAS/AIAA Astrodynamics Specialist Conference, 2024, pp. 1–19

work page 2024

[21] [21]

Dis- tributed Covariance Steering with Consensus ADMM for Stochastic Multi-Agent Systems,

A. D. Saravanos, A. Tsolovikos, E. Bakolas, and E. Theodorou, “Dis- tributed Covariance Steering with Consensus ADMM for Stochastic Multi-Agent Systems,” inProceedings of Robotics: Science and Systems, Virtual, July 2021

work page 2021

[22] [22]

Density control of interacting agent systems,

Y . Chen, “Density control of interacting agent systems,”IEEE Transac- tions on Automatic Control, vol. 69, no. 1, pp. 246–260, 2024

work page 2024

[23] [23]

Steering large agent populations using mean-field schr ¨odinger bridges with gaussian mixture models,

G. Rapakoulias, A. R. Pedram, and P. Tsiotras, “Steering large agent populations using mean-field schr ¨odinger bridges with gaussian mixture models,”IEEE Control Systems Letters, 2025

work page 2025

[24] [24]

Decentralized probabilistic den- sity control of autonomous swarms with safety constraints,

N. Demir, U. Eren, and B. Ac ¸ıkmes ¸e, “Decentralized probabilistic den- sity control of autonomous swarms with safety constraints,”Autonomous Robots, vol. 39, no. 4, pp. 537–554, 2015

work page 2015

[25] [25]

Group steering: Approaches based on power moments,

G. Wu and A. Lindquist, “Group steering: Approaches based on power moments,”arXiv preprint arXiv:2211.13370, 2022

work page arXiv 2022

[26] [26]

Distributed model predictive covariance steering,

A. D. Saravanos, I. M. Balci, E. Bakolas, and E. A. Theodorou, “Distributed model predictive covariance steering,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024, pp. 5740–5747

work page 2024

[27] [27]

Distributed Hierarchical Dis- tribution Control for Very-Large-Scale Clustered Multi-Agent Systems,

A. D. Saravanos, Y . Li, and E. Theodorou, “Distributed Hierarchical Dis- tribution Control for Very-Large-Scale Clustered Multi-Agent Systems,” inProceedings of Robotics: Science and Systems, Daegu, Republic of Korea, July 2023

work page 2023

[28] [28]

Robust motion coordi- nation with covariance steering model predictive control in bandwidth limited scenarios,

H. Bai, H. Zhu, X. Zhao, H. Li, and Y . Wang, “Robust motion coordi- nation with covariance steering model predictive control in bandwidth limited scenarios,” in2024 IEEE International Conference on Robotics and Biomimetics (ROBIO), 2024, pp. 1809–1814

work page 2024

[29] [29]

Distributed optimization and statistical learning via the alternating direction method of multipliers,

S. Boyd, N. Parikh, E. Chu, B. Peleato, and J. Eckstein, “Distributed optimization and statistical learning via the alternating direction method of multipliers,”Foundations and Trends® in Machine Learning, vol. 3, no. 1, pp. 1–122, 2011

work page 2011

[30] [30]

Distributed differential dynamic programming architectures for large-scale multia- gent control,

A. D. Saravanos, Y . Aoyama, H. Zhu, and E. A. Theodorou, “Distributed differential dynamic programming architectures for large-scale multia- gent control,”IEEE Transactions on Robotics, vol. 39, no. 6, pp. 4387– 4407, 2023

work page 2023

[31] [31]

Distributed opti- mization methods for multi-robot systems: Part 1—a tutorial [tutorial],

O. Shorinwa, T. Halsted, J. Yu, and M. Schwager, “Distributed opti- mization methods for multi-robot systems: Part 1—a tutorial [tutorial],” IEEE Robotics & Automation Magazine, vol. 31, no. 3, pp. 121–138, 2024

work page 2024

[32] [32]

Scalable robust optimization for safe multi-agent control under unknown deterministic uncertainty,

A. T. Abdul, A. D. Saravanos, and E. A. Theodorou, “Scalable robust optimization for safe multi-agent control under unknown deterministic uncertainty,” in2025 American Control Conference (ACC). IEEE, 2025, pp. 3666–3673

work page 2025

[33] [33]

Global convergence of splitting methods for nonconvex composite optimization,

G. Li and T. K. Pong, “Global convergence of splitting methods for nonconvex composite optimization,”SIAM Journal on Optimization, vol. 25, no. 4, pp. 2434–2460, 2015

work page 2015

[34] [34]

Convergence of alternating direction method for minimizing sum of two nonconvex functions with linear constraints,

K. Guo, D. Han, and T.-T. Wu, “Convergence of alternating direction method for minimizing sum of two nonconvex functions with linear constraints,”International Journal of Computer Mathematics, vol. 94, no. 8, pp. 1653–1669, 2017

work page 2017

[35] [35]

Convergence analysis of alternating direction method of multipliers for a family of nonconvex problems,

M. Hong, Z.-Q. Luo, and M. Razaviyayn, “Convergence analysis of alternating direction method of multipliers for a family of nonconvex problems,”SIAM Journal on Optimization, vol. 26, no. 1, pp. 337–364, 2016

work page 2016

[36] [36]

Linearized ADMM for nonconvex non- smooth optimization with convergence analysis,

Q. Liu, X. Shen, and Y . Gu, “Linearized ADMM for nonconvex non- smooth optimization with convergence analysis,”IEEE access, vol. 7, pp. 76 131–76 144, 2019

work page 2019

[37] [37]

Linearized ADMM converges to second-order stationary points for non-convex problems,

S. Lu, J. D. Lee, M. Razaviyayn, and M. Hong, “Linearized ADMM converges to second-order stationary points for non-convex problems,” IEEE Transactions on Signal Processing, vol. 69, pp. 4859–4874, 2021

work page 2021

[38] [38]

Iteration-complexity of a linearized proximal multiblock ADMM class for linearly constrained nonconvex optimization problems,

J. G. Melo and R. Monteiro, “Iteration-complexity of a linearized proximal multiblock ADMM class for linearly constrained nonconvex optimization problems,”Available on: http://www. optimization-online. org, 2017

work page 2017

[39] [39]

Global convergence of ADMM in nonconvex nonsmooth optimization,

Y . Wang, W. Yin, and J. Zeng, “Global convergence of ADMM in nonconvex nonsmooth optimization,”Journal of Scientific Computing, vol. 78, pp. 29–63, 2019

work page 2019

[40] [40]

Douglas–rachford splitting and admm for nonconvex optimization: Tight convergence results,

A. Themelis and P. Patrinos, “Douglas–rachford splitting and admm for nonconvex optimization: Tight convergence results,”SIAM Journal on Optimization, vol. 30, no. 1, pp. 149–181, 2020

work page 2020

[41] [41]

A two-level distributed algorithm for nonconvex constrained optimization,

K. Sun and X. A. Sun, “A two-level distributed algorithm for nonconvex constrained optimization,”Computational Optimization and Applica- tions, vol. 84, no. 2, pp. 609–649, 2023

work page 2023

[42] [42]

A two-level ADMM algorithm for AC OPF with global conver- gence guarantees,

——, “A two-level ADMM algorithm for AC OPF with global conver- gence guarantees,”IEEE Transactions on Power Systems, vol. 36, no. 6, pp. 5271–5281, 2021

work page 2021

[43] [43]

Fast and stable nonconvex constrained distributed optimization: the ELLADA algorithm,

W. Tang and P. Daoutidis, “Fast and stable nonconvex constrained distributed optimization: the ELLADA algorithm,”Optimization and Engineering, vol. 23, no. 1, pp. 259–301, 2022

work page 2022

[44] [44]

Constrained minimum variance and covariance steering based on affine disturbance feedback control pa- rameterization,

I. M. Balci and E. Bakolas, “Constrained minimum variance and covariance steering based on affine disturbance feedback control pa- rameterization,”International Journal of Robust and Nonlinear Control, vol. 34, no. 11, pp. 7332–7370, 2024

work page 2024

[45] [45]

Optimal stochastic vehicle path planning using covariance steering,

K. Okamoto and P. Tsiotras, “Optimal stochastic vehicle path planning using covariance steering,”IEEE Robotics and Automation Letters, vol. 4, no. 3, pp. 2276–2281, 2019

work page 2019

[46] [46]

Discrete-time optimal covariance steering via semidefinite programming,

G. Rapakoulias and P. Tsiotras, “Discrete-time optimal covariance steering via semidefinite programming,” in2023 62nd IEEE Conference on Decision and Control (CDC), 2023, pp. 1802–1807

work page 2023

[47] [47]

Nocedal and S

J. Nocedal and S. J. Wright,Numerical optimization. Springer, 2006

work page 2006

[48] [48]

Sequential quadratic programming,

P. T. Boggs and J. W. Tolle, “Sequential quadratic programming,”Acta numerica, vol. 4, pp. 1–51, 1995

work page 1995

[49] [49]

On the global and linear convergence of the generalized alternating direction method of multipliers,

W. Deng and W. Yin, “On the global and linear convergence of the generalized alternating direction method of multipliers,”Journal of Scientific Computing, vol. 66, pp. 889–916, 2016

work page 2016

[50] [50]

ApS,The MOSEK optimization toolbox for MATLAB manual

M. ApS,The MOSEK optimization toolbox for MATLAB manual. Version 9.0., 2019

work page 2019

[51] [51]

OSQP: An operator splitting solver for quadratic programs,

B. Stellato, G. Banjac, P. Goulart, A. Bemporad, and S. Boyd, “OSQP: An operator splitting solver for quadratic programs,”Mathematical Programming Computation, vol. 12, no. 4, pp. 637–672, 2020

work page 2020

[52] [52]

Gcbf+: A neural graph control barrier function framework for distributed safe multi-agent control,

S. Zhang, O. So, K. Garg, and C. Fan, “Gcbf+: A neural graph control barrier function framework for distributed safe multi-agent control,” IEEE Transactions on Robotics, 2025. SARAVANOSet al.: DISTRIBUTED COVARIANCE STEERING VIA NON-CONVEX ADMM FOR LARGE-SCALE MULTI-AGENT SYSTEMS 17

work page 2025

[53] [53]

Deep distributed optimization for large-scale quadratic programming,

A. D. Saravanos, H. Kuperman, A. Oshin, A. T. Abdul, V . Pacelli, and E. Theodorou, “Deep distributed optimization for large-scale quadratic programming,” inThe Thirteenth International Conference on Learning Representations, 2025. Augustinos D. Saravanos(Graduate Student Member, IEEE) received his Diploma in Elec- trical and Computer Engineering with hig...

work page 2025

[54] [54]

In 2011, he graduated with his PhD in Computer Science from USC

He also holds three MSc degrees in Production Engineering from TUC in 2003, Computer Science and Engineering from University of Minnesota in 2007, and Electrical Engineering from the University of Southern California (USC) in 2010. In 2011, he graduated with his PhD in Computer Science from USC. From 2011 to 2013, he was a Postdoc- toral Research Fellow w...

work page 2003