Arash Bahari Kordabad
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Arash Bahari Kordabad has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Sebastien Gros
NTNU
9 shared publications - Mario Zanon
IMT School for Advanced Studies Lucca
2 shared publications Sebastien Gros
2 shared publicationsWenqi Cai
2 shared publicationsAmirreza Valaei
1 shared publication- Dimos V. Dimarogonas
Department of Decision and Control Systems, School of Electrical Engineering and Computer Science, KTH Royal Institute of Technology, Stockholm, Sweden
1 shared publication
Latest Publications
- Open Access
- 28 September 2026
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 184
Second-order policy gradient methods for the linear quadratic regulator
- 28 September 2026
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 184
Open AccessSecond-order policy gradient methods for the linear quadratic regulator
- 2 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Control Systems Letters
- Vol. 10, 505-510
Multi-Agent Temporal Logic Planning via Penalty Functions and Block-Coordinate Optimization
Multi-agent planning under Signal Temporal Logic (STL) is often hindered by collaborative tasks that lead to computational challenges due to the inherent high dimensionality of the problem, preventing scalable synthesis with satisfaction guarantees. To address this, we formulate STL planning as an optimization program under multi-agent STL constraints and introduce a penalty-based unconstrained relaxation that can be efficiently solved via a Block-Coordinate Gradient Descent (BCGD) method, where each block corresponds to a single agent’s decision variables, thereby mitigating complexity. By utilizing a quadratic penalty function defined via smooth STL semantics, we show that BCGD iterations converge to a stationary point of the penalized problem under standard regularity assumptions. To enforce feasibility, the BCGD solver is embedded within a two-layer optimization scheme: inner BCGD updates are performed for a fixed penalty parameter, which is then increased in an outer loop to progressively improve multi-agent STL robustness. The proposed framework enables scalable computations and is validated through various complex multi-robot planning scenarios.
- 2 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Control Systems Letters
- Vol. 10, 505-510
Multi-Agent Temporal Logic Planning via Penalty Functions and Block-Coordinate Optimization
Multi-agent planning under Signal Temporal Logic (STL) is often hindered by collaborative tasks that lead to computational challenges due to the inherent high dimensionality of the problem, preventing scalable synthesis with satisfaction guarantees. To address this, we formulate STL planning as an optimization program under multi-agent STL constraints and introduce a penalty-based unconstrained relaxation that can be efficiently solved via a Block-Coordinate Gradient Descent (BCGD) method, where each block corresponds to a single agent’s decision variables, thereby mitigating complexity. By utilizing a quadratic penalty function defined via smooth STL semantics, we show that BCGD iterations converge to a stationary point of the penalized problem under standard regularity assumptions. To enforce feasibility, the BCGD solver is embedded within a two-layer optimization scheme: inner BCGD updates are performed for a fixed penalty parameter, which is then increased in an outer loop to progressively improve multi-agent STL robustness. The proposed framework enables scalable computations and is validated through various complex multi-robot planning scenarios.
- Open Access
- 21 June 2023
- Published by Elsevier in European Journal of Control
- Vol. 73
Learning-based state estimation and control using MHE and MPC schemes with imperfect models
- 21 June 2023
- Published by Elsevier in European Journal of Control
- Vol. 73
Open AccessLearning-based state estimation and control using MHE and MPC schemes with imperfect models
- 13 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Bias Correction of Discounted Optimal Steady-State using Cost Modification
In the literature of Economic Model Predictive Control (EMPC) and undiscounted Optimal Control Problem (OCP), the optimal steady-state point is an equilibrium point with the minimum stage cost. If the Economic MPC is discounted, this property does not hold, and the optimal steady-state point is not the same as the one obtained from the undiscounted EMPC. Therefore the discounted steady-state point does not yield minimum stage cost and has a bias with respect to the undiscounted one. In this paper, we propose a computationally inexpensive cost modification in the discounted MPC that results in the undiscounted optimal steady-state point, i.e., the steady-state point that leads to the best stage cost. Moreover, we show that this modification does not affect the closed-loop system behavior. We illustrate the proposed method in a numerical example.
- 13 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Bias Correction of Discounted Optimal Steady-State using Cost Modification
In the literature of Economic Model Predictive Control (EMPC) and undiscounted Optimal Control Problem (OCP), the optimal steady-state point is an equilibrium point with the minimum stage cost. If the Economic MPC is discounted, this property does not hold, and the optimal steady-state point is not the same as the one obtained from the undiscounted EMPC. Therefore the discounted steady-state point does not yield minimum stage cost and has a bias with respect to the undiscounted one. In this paper, we propose a computationally inexpensive cost modification in the discounted MPC that results in the undiscounted optimal steady-state point, i.e., the steady-state point that leads to the best stage cost. Moreover, we show that this modification does not affect the closed-loop system behavior. We illustrate the proposed method in a numerical example.
- 13 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Continuous-time Chance-constrained Stochastic Model Predictive Control using Multiple Shooting and CVaR
In this paper, we address the continuous-time chance-constrained nonlinear Stochastic Model Predictive Control (SMPC) problem using Conditional Value at Risk (CVaR). Chance constraint directly enforces a user-specified upper bound for the probability of constraint violation. However, it is known that a chance constraint may be non-convex and intractable in general. CVaR is a risk measure that provides a convex approximation of chance constraints. In this paper, we first use the multiple shooting method for the stochastic systems in order to represent a continuous-time Optimal Control Problem (OCP) in a Nonlinear Programming (NLP) form. Then we use scenario tree sampling to deal with the disturbances. Wheeled Mobile Robot (WMR) path planning with obstacle avoidance will be considered to illustrate the efficiency of the method.
- 13 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Continuous-time Chance-constrained Stochastic Model Predictive Control using Multiple Shooting and CVaR
In this paper, we address the continuous-time chance-constrained nonlinear Stochastic Model Predictive Control (SMPC) problem using Conditional Value at Risk (CVaR). Chance constraint directly enforces a user-specified upper bound for the probability of constraint violation. However, it is known that a chance constraint may be non-convex and intractable in general. CVaR is a risk measure that provides a convex approximation of chance constraints. In this paper, we first use the multiple shooting method for the stochastic systems in order to represent a continuous-time Optimal Control Problem (OCP) in a Nonlinear Programming (NLP) form. Then we use scenario tree sampling to deal with the disturbances. Wheeled Mobile Robot (WMR) path planning with obstacle avoidance will be considered to illustrate the efficiency of the method.
- Open Access
- 27 May 2023
- Published by Springer Nature in International Journal of Dynamics and Control
- Vol. 12 (3) , 878-890
Lyapunov-based robust optimal control for time-delay systems with application in milling process
In this paper, we propose an optimal delay-independent robust controller based on the Lyapunov–Krakovskii theorem for the milling process as a time-delay system in the presence of varying axial depth of cut and different parametric uncertainty, such as stiffness, damping, etc. Milling is widely used in the manufacturing processes for the production of complex-shaped workpieces with high accuracy. However, chatter is a self-excited vibration that may cause adverse effects, such as tool damage, poor surface quality, excessive noise, etc. The dynamic model of the milling process is considered a two-dimensional time-delay system. A nonlinear programming with linear matrix inequality constraints is solved in order to obtain the controller gain where the objective function is corresponding to the norm-2 of controller gain, and the constraints guarantee robust stability. Using the semi-discretization method, stability lobes are shown in both uncontrolled and controlled plants to illustrate the improvement of the stable region via the proposed controller. Bifurcation phenomena have been improved with this controller by postponing the adverse effects to the higher values of the axial depth of cut, reducing the amplitude of limit cycles, and changing the type of bifurcation. Finally, we will compare the proposed controller with an intelligent controller in order to show the efficiency of the proposed method. It is shown that the proposed controller has improved the integral absolute error index by about 3.65 times compared with the intelligent controller.
- 27 May 2023
- Published by Springer Nature in International Journal of Dynamics and Control
- Vol. 12 (3) , 878-890
Open AccessLyapunov-based robust optimal control for time-delay systems with application in milling process
In this paper, we propose an optimal delay-independent robust controller based on the Lyapunov–Krakovskii theorem for the milling process as a time-delay system in the presence of varying axial depth of cut and different parametric uncertainty, such as stiffness, damping, etc. Milling is widely used in the manufacturing processes for the production of complex-shaped workpieces with high accuracy. However, chatter is a self-excited vibration that may cause adverse effects, such as tool damage, poor surface quality, excessive noise, etc. The dynamic model of the milling process is considered a two-dimensional time-delay system. A nonlinear programming with linear matrix inequality constraints is solved in order to obtain the controller gain where the objective function is corresponding to the norm-2 of controller gain, and the constraints guarantee robust stability. Using the semi-discretization method, stability lobes are shown in both uncontrolled and controlled plants to illustrate the improvement of the stable region via the proposed controller. Bifurcation phenomena have been improved with this controller by postponing the adverse effects to the higher values of the axial depth of cut, reducing the amplitude of limit cycles, and changing the type of bifurcation. Finally, we will compare the proposed controller with an intelligent controller in order to show the efficiency of the proposed method. It is shown that the proposed controller has improved the integral absolute error index by about 3.65 times compared with the intelligent controller.
- 17 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automatic Control
- Vol. 69 (2) , 1149-1156
Equivalence of Optimality Criteria for Markov Decision Process and Model Predictive Control
This paper shows that the optimal policy and value functions of a Markov Decision Process (MDP), either discounted or not, can be captured by a finite-horizon undiscounted Optimal Control Problem (OCP), even if based on an inexact model. This can be achieved by selecting a proper stage cost and terminal cost for the OCP. A very useful particular case of OCP is a Model Predictive Control (MPC) scheme where a deterministic (possibly nonlinear) model is used to reduce the computational complexity. This observation leads us to parameterize an MPC scheme fully, including the cost function. In practice, Reinforcement Learning algorithms can then be used to tune the parameterized MPC scheme. We verify the developed theorems analytically in an LQR case and we investigate some other nonlinear examples in simulations.
- 17 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automatic Control
- Vol. 69 (2) , 1149-1156
Equivalence of Optimality Criteria for Markov Decision Process and Model Predictive Control
This paper shows that the optimal policy and value functions of a Markov Decision Process (MDP), either discounted or not, can be captured by a finite-horizon undiscounted Optimal Control Problem (OCP), even if based on an inexact model. This can be achieved by selecting a proper stage cost and terminal cost for the OCP. A very useful particular case of OCP is a Model Predictive Control (MPC) scheme where a deterministic (possibly nonlinear) model is used to reduce the computational complexity. This observation leads us to parameterize an MPC scheme fully, including the cost function. In practice, Reinforcement Learning algorithms can then be used to tune the parameterized MPC scheme. We verify the developed theorems analytically in an LQR case and we investigate some other nonlinear examples in simulations.
- Open Access
Equivalence of Optimality Criteria for Markov Decision Process and Model Predictive Control
This paper shows that the optimal policy and value functions of a Markov Decision Process (MDP), either discounted or not, can be captured by a finite-horizon undiscounted Optimal Control Problem (OCP), even if based on an inexact model. This can be achieved by selecting a proper stage cost and terminal cost for the OCP. A very useful particular case of OCP is a Model Predictive Control (MPC) scheme where a deterministic (possibly nonlinear) model is used to reduce the computational complexity. This observation leads us to parameterize an MPC scheme fully, including the cost function. In practice, Reinforcement Learning algorithms can then be used to tune the parameterized MPC scheme. We verify the developed theorems analytically in an LQR case and we investigate some other nonlinear examples in simulations.
Open AccessEquivalence of Optimality Criteria for Markov Decision Process and Model Predictive Control
This paper shows that the optimal policy and value functions of a Markov Decision Process (MDP), either discounted or not, can be captured by a finite-horizon undiscounted Optimal Control Problem (OCP), even if based on an inexact model. This can be achieved by selecting a proper stage cost and terminal cost for the OCP. A very useful particular case of OCP is a Model Predictive Control (MPC) scheme where a deterministic (possibly nonlinear) model is used to reduce the computational complexity. This observation leads us to parameterize an MPC scheme fully, including the cost function. In practice, Reinforcement Learning algorithms can then be used to tune the parameterized MPC scheme. We verify the developed theorems analytically in an LQR case and we investigate some other nonlinear examples in simulations.
- Top Cited PapersOpen Access
- 6 January 2023
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 119
Energy management in residential microgrid using model predictive control-based reinforcement learning and Shapley value
- 6 January 2023
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 119
Top Cited PapersOpen AccessEnergy management in residential microgrid using model predictive control-based reinforcement learning and Shapley value
- Open Access
- 12 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 130058-130067
Safe Reinforcement Learning Using Wasserstein Distributionally Robust MPC and Chance Constraint
In this paper, we address the chance-constrained safe Reinforcement Learning (RL) problem using the function approximators based on Stochastic Model Predictive Control (SMPC) and Distributionally Robust Model Predictive Control (DRMPC). We use Conditional Value at Risk (CVaR) to measure the probability of constraint violation and safety. In order to provide a safe policy by construction, we first propose using parameterized nonlinear DRMPC at each time step. DRMPC optimizes a finite-horizon cost function subject to the worst-case constraint violation in an ambiguity set. We use a statistical ball around the empirical distribution with radius measured by the Wasserstein metric as the ambiguity set. Unlike the sample average approximation SMPC, DRMPC provides a probabilistic guarantee of the out of sample risk and requires lower samples from the disturbance. Then the Q-learning method is used to optimize the parameters in the DRMPC to achieve the best closed-loop performance. Wheeled Mobile Robot (WMR) path planning with obstacle avoidance will be considered to illustrate the efficiency of the proposed method.
- 12 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 130058-130067
Open AccessSafe Reinforcement Learning Using Wasserstein Distributionally Robust MPC and Chance Constraint
In this paper, we address the chance-constrained safe Reinforcement Learning (RL) problem using the function approximators based on Stochastic Model Predictive Control (SMPC) and Distributionally Robust Model Predictive Control (DRMPC). We use Conditional Value at Risk (CVaR) to measure the probability of constraint violation and safety. In order to provide a safe policy by construction, we first propose using parameterized nonlinear DRMPC at each time step. DRMPC optimizes a finite-horizon cost function subject to the worst-case constraint violation in an ambiguity set. We use a statistical ball around the empirical distribution with radius measured by the Wasserstein metric as the ambiguity set. Unlike the sample average approximation SMPC, DRMPC provides a probabilistic guarantee of the out of sample risk and requires lower samples from the disturbance. Then the Q-learning method is used to optimize the parameters in the DRMPC to achieve the best closed-loop performance. Wheeled Mobile Robot (WMR) path planning with obstacle avoidance will be considered to illustrate the efficiency of the proposed method.