Pith. sign in

REVIEW 3 major objections 5 minor 42 references

Toward Near-Globally Optimal Nonlinear Model Predictive Control via Diffusion Models

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read A diffusion model trained on local optima yields a near-globally optimal NMPC policy with high probability.

desk verdict A plausible diffusion-based NMPC pipeline whose near-global guarantee rests on an unverified positive-probability assumption about the data-generating process. read the letter →

arxiv 2412.08278 v3 pith:G56SE3LS submitted 2024-12-11 eess.SY cs.SY

classification eess.SYcs.SY
keywords diffusionmodelnonlinearpredictivecontrolglobaloptimizationrandomshootingprobabilisticnear-globaloptimalitymultimodaldistributionlearning-basedsample-score-rank
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Nonlinear model predictive control (NMPC) forces a trade-off: fast local solvers get stuck in poor local optima, while global solvers are too slow for real-time use. This paper argues that a diffusion model can break the trade-off by learning the multi-modal distribution of optimal control sequences. Offline, it runs a cheap local optimizer from many random initial guesses to generate a dataset of candidate solutions; online, it samples a batch of candidate sequences from the trained model, evaluates their costs, and applies the best one. The central result is a probabilistic guarantee: with $M$ independent samples and a diffusion model that is within $\tilde{\delta}$ of the true solution distribution, the chance that the selected sequence misses the $\varepsilon$-ball around the global optimum is at most $(1-p_B+\tilde{\delta})^M$, which shrinks exponentially as $M$ grows. The simulations on cart-pole, pendubot, and double cart-pole show costs close to the global optimizer at computation times comparable to a local solver.

What carries the argument

The mechanism that carries the argument is a conditional diffusion model $p_\theta(\cdot|x_t)$, trained with classifier-free guidance to approximate the multi-modal distribution $\rho(\cdot|x_t)$ of optimal control sequences that a local NMPC solver produces from random initial guesses. The online phase, Algorithm 2, is a random-shooting loop: sample $M$ sequences from $p_\theta$, evaluate the OCP cost $J(x_t,\cdot)$ for each, and return the sequence with the lowest cost. The load-bearing identity is the error bound $\delta=(1-p_B+\tilde{\delta})^M$ from Theorem 3, which converts the positive event that a single sample lands in the $\varepsilon$-ball around the global optimum into an exponentially reliable selection guarantee as $M$ increases.

What would settle it

Take any benchmark OCP with a known global optimum $u^*$, sample $N$ random initial guesses from the full support of $\phi$, run the local solver on each, and count the fraction $p_B$ of runs that land within $\varepsilon$ of $u^*$. If $p_B=0$ for a state in the reachable set while the diffusion model is trained on that state, then Theorem 2's premise is violated and Algorithm 2 cannot be near-globally optimal; conversely, if $p_B>0$, the bound $\delta=(1-p_B+\tilde{\delta})^M$ gives a concrete, testable prediction for the reliability as a function of $M$ and the estimated model error $\tilde{\delta}$.

Watch

Extended reading notes

Core claim

On the paper's own terms, the claim is that the distribution of locally optimal NMPC solutions -- obtained by applying a local optimizer to uniformly random initial guesses -- is learnable by a diffusion model, and that sampling from that learned distribution in a sample-score-rank loop is probabilistically near-globally optimal. Theorem 3 states this precisely: if the diffusion model approximates the true solution distribution $\rho(\cdot|x_t)$ to within an error $\tilde{\delta}$ on every $\varepsilon$-ball, and if $\tilde{\delta}$ is smaller than $p_B$, the probability mass that $\rho$ puts on the $\varepsilon$-ball around the global optimum $u^*_{[0:H]}(x_t)$, then drawing $M$ independent candidate sequences and keeping the cheapest one satisfies $\mathbb{P}(\|u-u^*_{[0:H]}(x_t)\|_2\leq\varepsilon)\geq 1-\delta$ with $\delta=(1-p_B+\tilde{\delta})^M$. Because $\delta$ decays exponentially in $M$, arbitrarily high reliability is achieved by drawing more samples, which is cheap on parallel hardware. The paper also proves (Theorem 2) that, under a positive-probability condition on reaching the global basin from random initial guesses, the offline dataset asymptotically contains dense global optima, so the approach avoids the covariance-shift problem that plagues direct behavior cloning.

Load-bearing premise

That for every state the system can reach, a randomly initialized local solver has some chance of finding the absolute best control sequence, not just a locally good one.

Editorial extensions

If this is right

  • Real-time NMPC on systems with fast dynamics could reach global-optimizer-level performance without any online global optimization, since the diffusion model's forward pass is GPU-parallelizable and the reliability is tuned purely by the sample count $M$.
  • The learned distribution can serve as a high-quality warm start for local solvers, combining the probabilistic guarantee with exact constraint satisfaction in an online refinement stage.
  • For systems with discrete switching between locally optimal behaviours (e.g., left and right swing-up of a cart-pole), the multimodal learned distribution naturally represents both strategies, so the controller can switch between them as the state evolves.
  • Because $\delta$ decays exponentially in $M$, the computational budget for a desired reliability level is directly predictable from the model quality $\tilde{\delta}$ and the probability mass $p_B$ around the global optimum.
  • The framework reduces global NMPC to a supervised learning problem: any improvement in diffusion model accuracy lowers the required sample count $M$ for a fixed $\delta$.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The guarantee is only as strong as Theorem 2's implicit assumption that the local solver reaches the global basin with positive probability; if the global optimum's basin of attraction has zero measure in the space of initial guesses, the dataset will never contain it and the near-global claim collapses regardless of $M$.
  • The paper defers the proofs of Theorems 2 and 3 to an extended version, so the formal guarantee must be verified in that document; the present manuscript states the results without their derivations.
  • A testable extension is to compute the empirical miss rate of Algorithm 2 on a benchmark with a known global optimum and compare it to the bound $(1-p_B+\tilde{\delta})^M$; a systematic violation would indicate that $\tilde{\delta}$ underestimates the model error.
  • The method is essentially replacing multistart global optimization with a learned proposal distribution; it could be combined with importance sampling or sequential Monte Carlo to obtain unbiased estimates of the global optimum rather than only a bound.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a diffusion-model-based approximation of globally optimal nonlinear model predictive control (NMPC). Offline, Algorithm 1 generates a dataset by applying a local optimizer to random initial guesses, producing a multimodal distribution rho(·|x_t) of locally optimal control sequences, and trains a conditional diffusion model p_theta to approximate rho. Online, Algorithm 2 draws M control sequences from the diffusion model, evaluates their costs, and selects the lowest-cost sequence. The central theoretical result, Theorem 3, states that if the diffusion approximation error satisfies (10) with tilde_delta < p_B, where p_B is the rho-mass of an epsilon-ball around the global optimum, then Algorithm 2 is probabilistically near-globally optimal with delta = (1 - p_B + tilde_delta)^M. The numerical evaluation on cart-pole, pendubot, and double cart-pole reports lower costs than direct neural-network approximations and computation times comparable to or lower than local NMPC, while matching the cost of a global-optimization baseline.

Significance. If the stated assumptions could be verified, the result would be significant for NMPC practice: it would show that a diffusion model trained on locally optimal solutions can replace expensive online global optimization by parallelizable sampling, preserving a probabilistic near-global optimality guarantee. The paper's strengths include a clean probabilistic formulation of near-global optimality, a practical offline data-generation scheme, a clear comparison against local and global NMPC and neural-network baselines, and ablation-style experiments varying M, H, and K. The main weakness is that the two load-bearing hypotheses of Theorem 3—positive probability p_B of hitting the global optimum under the local-solver-induced distribution, and a certified approximation error tilde_delta < p_B—are not verified either theoretically or empirically. Section 5 explicitly defers formal theoretical optimality guarantees to future work, which is consistent with the conditional nature of the current result. The manuscript therefore presents a promising empirical method and a conditional guarantee, but the near-global optimality claim is not yet established for the tested or general problems.

major comments (3)
  1. [Section 3.1, Theorem 2] The assumption P(u*_{[0:H]}(x)) > 0 for all x is load-bearing but not justified by the argument given. The local optimizer S is a deterministic operator for each fixed state x; under an initial-guess distribution phi, the probability of reaching the global optimum is the phi-measure of the basin of attraction of u*_{[0:H]}(x) under S. A uniform phi with positive support on U^H only ensures that initial guesses are drawn densely; it does not ensure that this basin has positive volume. For OCPs in which a global optimum lies on the boundary of U^H, is a degenerate stationary point, or exists only at a bifurcation value of the state, the basin can have measure zero, giving P(u*_{[0:H]}(x)) = 0. In that case the dataset generated by Algorithm 1 contains no samples near the global optimum for those states, the condition p_B > 0 in Theorem 3 fails, and no number M of online samples can compensate. The paper should either prove a sufficient condition for positive basin volume for the considered problem class, restrict the claim to states where this has been verified, or provide empirical estimates of p_B for all states visited in the simulations.
  2. [Section 3.2, Eq. (10) and Theorem 3] Equation (10) is asserted to be achievable 'given a sufficiently large data set D' and a sufficiently flexible model, but the paper provides no sample-complexity bound, no training guarantee, and no verification that the trained diffusion model satisfies tilde_delta < p_B for every x in X and every epsilon-ball around the global optimum. Since p_B can be extremely small in the high-dimensional control-sequence space U^H, the condition tilde_delta < p_B is nontrivial and likely difficult to certify in practice. The theorem's conclusion delta = (1 - p_B + tilde_delta)^M depends critically on this strict inequality; without a way to estimate p_B or bound tilde_delta, the near-global optimality guarantee cannot be instantiated for any of the numerical examples. The paper should state how p_B and tilde_delta would be estimated or bounded for the considered systems, or explicitly restrict the theorem to cases where such bounds are available.
  3. [Section 5] The conclusion states that 'future work will focus on deriving formal theoretical optimality guarantees for our proposed approach.' This admission is in tension with Theorem 3, which is presented as a probabilistic near-global optimality guarantee. The tension is resolved if Theorem 3 is understood as a conditional guarantee whose assumptions are not verified, but the manuscript should say this explicitly and should specify which parts of the optimality claim are currently formal and which remain heuristic. As written, a reader could reasonably conclude that the near-global optimality of Algorithm 2 has been established for the tested systems when in fact only a conditional statement with unverified hypotheses is provided.
minor comments (5)
  1. [Algorithm 1, line 9] There is a mismatch between the algorithm and the text: Algorithm 1 propagates the system using x^d_t, i.e., x_{t+1} <- f(x^d_t, \hat{u}_0(x^d_t)), while the surrounding text says 'Apply \hat{u}_0(x_t) to system (1): x_{t+1} <- f(x_t, \hat{u}_0(x_t)).' Please clarify which state is used for the roll-out, as this affects the state distribution in the dataset.
  2. [Section 2, Eq. (3)] The stochastic policy is defined as pi : R^{n_u} x R^{n_x} -> R_{>=0}, but in (3) it is used as a distribution over control sequences u in U^H conditioned on x_t. The domain should be U^H x X or the notation should be adjusted to match the conditional density interpretation.
  3. [Theorem 3] The expression delta = (1 - p_B + tilde_delta)^M is only guaranteed to be a probability in [0,1] when the condition tilde_delta < p_B holds. This is stated, but the proof sketch could make clearer that this inequality is used both to ensure the base of the exponent is less than 1 and to ensure the Markov bound is meaningful.
  4. [Section 4.1] The text refers to 'CASADOS (Verschueren et al., 2021)' but the cited software is ACADOS. This looks like a typo and should be corrected.
  5. [Figures 2 and 3] The cost and computation-time comparisons are shown without error bars or confidence intervals. Since the method is probabilistic and the baselines involve random initializations, reporting variability across seeds or initial states would strengthen the empirical claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the main theorem is a conditional probability bound with an explicit assumption (p_B), and the numerical evaluation gives independent empirical support.

full rationale

The claimed derivation is conditional rather than circular. Section 3.1 defines the data distribution rho by pushing a uniform initial-guess distribution phi through the local optimizer S, and Algorithm 1 samples (x, u-hat) pairs from it. Theorem 2 states that, under P(u*[0:H](x))>0 and positive state-perturbation variance, global optima eventually appear in the dataset; this is a coverage assertion about the sampling scheme, not an output that is later renamed as a prediction. Theorem 3 then takes the diffusion-model approximation error (10) and the rho-mass p_B as hypotheses and derives the random-shooting bound delta=(1-p_B+tilde_delta)^M; p_B is not fitted from the dataset and the bound is a direct union-bound arithmetic fact. The online phase (Algorithm 2) scores samples by the original cost J, so the selection step is evaluated on fresh costs and is not forced by a trained parameter. The paper's only self-citation is a pointer to the extended version for proofs, which is not a load-bearing appeal to a uniqueness theorem or an ansatz. The concern that p_B > 0 may fail for some nonconvex OCPs is a correctness/assumption risk about the data-generating process, not circularity; the paper explicitly states the assumption and the theorem is conditional on it. Therefore no circular step can be exhibited.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The method relies on a fitted diffusion model and several hand-chosen hyperparameters, plus assumptions about the solver's ability to reach the global optimum and the model's approximation accuracy. No new physical entities are introduced.

free parameters (5)
  • Diffusion model parameters theta = not reported, thousands of weights
    Trained by minimizing Eq. (9) on the generated dataset; the online sampler is defined by these parameters, so the entire result depends on this fit.
  • Number of online samples M = 20 in Section 4.2, varied in Fig. 3
    Hand-chosen; directly controls the guarantee delta = (1 - p_B + ~delta)^M, but no procedure is given to select M for a target delta.
  • Initial guess distribution phi = uniform over [-u_{t-1}, u_{t-1}] per Section 4.1
    Hand-chosen support; determines whether the local solver finds the global optimum during data generation, which is the crux of the near-global optimality claim.
  • Perturbation variance schedule sigma(t) = not specified precisely
    Hand-chosen in Algorithm 1; controls state-space coverage and covariance-shift mitigation; the paper does not report the schedule.
  • Diffusion hyperparameters (K, beta_k, p) = p = 0.25, K sensitivity explored in Fig. 3; beta schedule not specified
    Chosen without systematic tuning; the paper shows robustness to K only above a threshold.
assumptions (5)
  • ad hoc to paper P(u*[0:H](x)) > 0 for all x under the distribution induced by phi and the local solver S.
    Explicitly assumed in Theorem 2; the paper argues positive-support phi suffices, but positive support does not imply positive probability of the solver reaching the global basin. This premise is what turns local-optima samples into a near-global guarantee.
  • standard math The diffusion model can approximate rho(·|x_t) arbitrarily well given enough data and model capacity, as in Eq. (10).
    Invoked in Section 3.2 via Sohl-Dickstein et al. 2015; no finite-sample bound or network architecture satisfying it is given.
  • ad hoc to paper The model error satisfies ~delta < p_B for every state and every epsilon-ball around the global optimum.
    Assumed in Theorem 3; p_B is the mass of rho in the epsilon-ball. The paper gives no way to check or enforce this, so the advertised delta bound is conditional.
  • domain assumption Stage and terminal costs and dynamics are differentiable, and X and U are compact; no state constraints.
    Used in the OCP formulation in Section 2 and Remark 1 excludes state constraints, limiting the scope to unconstrained-input problems.
  • standard math State perturbations with Gaussian noise make the training-state distribution asymptotically dense in X.
    Used to argue robustness to covariance shift and coverage; true with positive variance, but the online guarantee for exact states relies on smoothness and interpolation that are not formalized.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Toward Near-Globally Optimal Nonlinear Model Predictive Control via Diffusion Models." pith.science (2026). https://pith.science/paper/G56SE3LS

@misc{pith2026241208278,
  author       = {Pith},
  title        = {Pith review of: Toward Near-Globally Optimal Nonlinear Model Predictive Control via Diffusion Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G56SE3LS}},
  note         = {Machine review of arXiv:2412.08278}
}
read the original abstract

Achieving global optimality in nonlinear model predictive control (NMPC) is challenging due to the non-convex nature of the underlying optimization problem. Since commonly employed local optimization techniques depend on carefully chosen initial guesses, this non-convexity often leads to suboptimal performance resulting from local optima. To overcome this limitation, we propose a novel diffusion model-based approach for near-globally optimal NMPC consisting of an offline and an online phase. The offline phase employs a local optimizer to sample from the distribution of optimal NMPC control sequences along generated system trajectories through random initial guesses. Subsequently, the generated diverse dataset is used to train a diffusion model to reflect the multi-modal distribution of optima. In the online phase, the trained model is leveraged to efficiently perform a variant of random shooting optimization to obtain near-globally optimal control sequences without relying on any initial guesses or online NMPC solving. The effectiveness of our approach is illustrated in a numerical simulation indicating high performance benefits compared to direct neural network approximations of NMPC and significantly lower computation times than online solving NMPC using global optimizers.

Figures

Figures reproduced from arXiv: 2412.08278 by the authors.

Figure 1
Figure 1. Left: Evolution of the pole angle θ resulting from two kinds of optimal control sequences, whose multimodal distribution is illustrated for different time steps. Right: Comparison of the multimodality of samples from our trained diffusion model and samples employed by a global optimizer for different time steps along roll-out trajectories. the left and right swing up, generally, at least two locally optimal solution… view at source ↗
Figure 2
Figure 2. Top: Our diffusion model-based approach achieves similar cost as a global optimization approach for NMPC, but significantly outperforms direct neural network approximations of NMPC. Bottom: The required computation times for online optimization based on a diffusion model are significantly reduced compared to the direct global optimization approach and at most on par with the commonly employed local optimizer. 4.2. R… view at source ↗
Figure 3
Figure 3. Performance of diffusion model-based and global optimization-based NMPC depending on the number of sampled control sequences M (left), horizon length H (middle), and diffusion steps K (right). alization. While our diffusion model approach is a significant improvement over the neural network approximation of the globally optimal NMPC scheme for all the investigated system dynamics, its complexity is comparable to or … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 30 canonical work pages

  1. [1]

    Tenenbaum, Tommi S

    Anurag Ajay, Yilun Du, Abhi Gupta, Joshua B. Tenenbaum, Tommi S. Jaakkola, and Pulkit Agrawal. Is conditional generative modeling all you need for decision making? In Proceedings of the International Conference on Learning Representations, 2023

  2. [2]

    Casadi: a software framework for nonlinear optimization and optimal control

    Joel AE Andersson, Joris Gillis, Greg Horn, James B Rawlings, and Moritz Diehl. Casadi: a software framework for nonlinear optimization and optimal control. Mathematical Programming Computation, 11: 0 1--36, 2019

  3. [3]

    The explicit linear quadratic regulator for constrained systems

    Alberto Bemporad, Manfred Morari, Vivek Dua, and Efstratios N Pistikopoulos. The explicit linear quadratic regulator for constrained systems. Automatica, 38 0 (1): 0 3--20, 2002

  4. [4]

    Bonzanini, Joel A

    Angelo D. Bonzanini, Joel A. Paulson, Georgios Makrygiorgos, and Ali Mesbah. Fast approximate learning-based multistage nonlinear model predictive control using gaussian processes and deep neural networks. Computers & Chemical Engineering, 145: 0 107174, 2021

  5. [5]

    Motion planning diffusion: Learning and planning of robot motions with diffusion models

    Joao Carvalho, An T Le, Mark Baierl, Dorothea Koert, and Jan Peters. Motion planning diffusion: Learning and planning of robot motions with diffusion models. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 1916--1923, 2023

  6. [6]

    Transformer-based model predictive control: Trajectory optimization via sequence modeling

    Davide Celestini, Daniele Gammelli, Tommaso Guffanti, Simone D'Amico, Elisa Capello, and Marco Pavone. Transformer-based model predictive control: Trajectory optimization via sequence modeling. IEEE Robotics and Automation Letters, 9 0 (11): 0 9820--9827, 2024

  7. [7]

    Diffusion policy: Visuomotor policy learning via action diffusion

    Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burchfiel, Russ Tedrake, and Shuran Song. Diffusion policy: Visuomotor policy learning via action diffusion. The International Journal of Robotics Research, page 02783649241273668, 2023

  8. [8]

    Recent theoretical advances in non-convex optimization

    Marina Danilova, Pavel Dvurechensky, Alexander Gasnikov, Eduard Gorbunov, Sergey Guminov, Dmitry Kamzolov, and Innokentiy Shibaev. Recent theoretical advances in non-convex optimization. In High-Dimensional Optimization and Probability: With a View Towards Data Science, pages 79--163. Springer, 2022

Show all 42 references
  1. [9]

    Model predictive control and its application in agriculture: A review

    Ying Ding, Liang Wang, Yongwei Li, and Daoliang Li. Model predictive control and its application in agriculture: A review. Computers and Electronics in Agriculture, 151: 0 104--117, 2018

  2. [10]

    Optimal control of a laboratory binary distillation column via regionless explicit mpc

    J \'a n Drgo n a, Martin Klau c o, Filip Jane c ek, and Michal Kvasnica. Optimal control of a laboratory binary distillation column via regionless explicit mpc. Computers & Chemical Engineering, 96: 0 139--148, 2017

  3. [11]

    One step diffusion via shortcut models

    Kevin Frans, Danijar Hafner, Sergey Levine, and Pieter Abbeel. One step diffusion via shortcut models. arXiv preprint arXiv:2410.12557, 2024

  4. [12]

    An implicit and explicit dual model predictive control formulation for a steel recycling process

    Andrea Ghezzi, Florian Messerer, Jacopo Balocco, Vincenzo Manzoni, and Moritz Diehl. An implicit and explicit dual model predictive control formulation for a steel recycling process. European Journal of Control, 74: 0 100841, 2023

  5. [13]

    Lane-merging using policy-based reinforcement learning and post-optimization

    Patrick Hart, Leonard Rychly, and Alois Knoll. Lane-merging using policy-based reinforcement learning and post-optimization. In 2019 IEEE Intelligent Transportation Systems Conference (ITSC), pages 3176--3181. IEEE, 2019

  6. [14]

    Keep it upright: Model predictive control for nonprehensile object transportation with obstacle avoidance on a mobile manipulator

    Adam Heins and Angela P Schoellig. Keep it upright: Model predictive control for nonprehensile object transportation with obstacle avoidance on a mobile manipulator. IEEE Robotics and Automation Letters, 2023

  7. [15]

    Learning an approximate model predictive controller with guarantees

    Michael Hertneck, Johannes Köhler, Sebastian Trimpe, and Frank Allgöwer. Learning an approximate model predictive controller with guarantees. IEEE Control Systems Letters, 2 0 (3): 0 543--548, 2018

  8. [16]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598, 2022

  9. [17]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. In Advances in Neural Information Processing Systems, volume 33, pages 6840--6851, 2020

  10. [18]

    Diffusion-based generation, optimization, and planning in 3d scenes

    Siyuan Huang, Zan Wang, Puhao Li, Baoxiong Jia, Tengyu Liu, Yixin Zhu, Wei Liang, and Song-Chun Zhu. Diffusion-based generation, optimization, and planning in 3d scenes. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16750--16761, 2023

  11. [19]

    Planning with diffusion for flexible behavior synthesis

    Michael Janner, Yilun Du, Joshua B Tenenbaum, and Sergey Levine. Planning with diffusion for flexible behavior synthesis. International Conference on Machine Learning, 2022

  12. [20]

    Efficient representation and approximation of model predictive control laws via deep learning

    Benjamin Karg and Sergio Lucia. Efficient representation and approximation of model predictive control laws via deep learning. IEEE Transactions on Cybernetics, 50 0 (9): 0 3866--3878, 2020

  13. [21]

    Efficient and guaranteed-safe non-convex trajectory optimization with constrained diffusion model

    Anjian Li, Zihan Ding, Adji Bousso Dieng, and Ryne Beeson. Efficient and guaranteed-safe non-convex trajectory optimization with constrained diffusion model. arXiv preprint arXiv:2403.05571, 2024

  14. [22]

    Nonlinear mpc for collision avoidance and control of uavs with dynamic obstacles

    Björn Lindqvist, Sina Sharif Mansouri, Ali-akbar Agha-mohammadi, and George Nikolakopoulos. Nonlinear mpc for collision avoidance and control of uavs with dynamic obstacles. IEEE Robotics and Automation Letters, 5 0 (4): 0 6001--6008, 2020

  15. [23]

    Safe and fast tracking on a robot manipulator: Robust mpc and neural network control

    Julian Nubert, Johannes Köhler, Vincent Berenz, Frank Allgöwer, and Sebastian Trimpe. Safe and fast tracking on a robot manipulator: Robust mpc and neural network control. IEEE Robotics and Automation Letters, 5 0 (2): 0 3050--3057, 2020

  16. [24]

    Jones, Dimitrios Gyalistras, Markus Gwerder, Vanessa Stauch, Beat Lehmann, and Manfred Morari

    Frauke Oldewurtel, Alessandra Parisio, Colin N. Jones, Dimitrios Gyalistras, Markus Gwerder, Vanessa Stauch, Beat Lehmann, and Manfred Morari. Use of model predictive control and weather forecasts for energy efficient building climate control. Energy and Buildings, 45: 0 15--27, 2012

  17. [25]

    Sample-efficient cross-entropy method for real-time planning

    Cristina Pinneri, Shambhuraj Sawant, Sebastian Blaes, Jan Achterhold, Joerg Stueckler, Michal Rolinek, and Georg Martius. Sample-efficient cross-entropy method for real-time planning. In Conference on Robot Learning, pages 1049--1065. PMLR, 2021

  18. [26]

    An overview of industrial model predictive control technology

    S Joe Qin and Thomas A Badgwell. An overview of industrial model predictive control technology. In AIche symposium series, volume 93, pages 232--256, 1997

  19. [27]

    Toward the fundamental limits of imitation learning

    Nived Rajaraman, Lin Yang, Jiantao Jiao, and Kannan Ramchandran. Toward the fundamental limits of imitation learning. Advances in Neural Information Processing Systems, 33: 0 2914--2924, 2020

  20. [28]

    Application of interior-point methods to model predictive control

    Christopher V Rao, Stephen J Wright, and James B Rawlings. Application of interior-point methods to model predictive control. Journal of optimization theory and applications, 99: 0 723--757, 1998

  21. [29]

    Model predictive control: theory, computation, and design, volume 2

    James Blake Rawlings, David Q Mayne, Moritz Diehl, et al. Model predictive control: theory, computation, and design, volume 2. Nob Hill Publishing Madison, WI, 2017

  22. [30]

    Ac4mpc: Actor-critic reinforcement learning for nonlinear model predictive control

    Rudolf Reiter, Andrea Ghezzi, Katrin Baumg \ A \=I rtner, Jasper Hoffmann, Robert D McAllister, and Moritz Diehl. Ac4mpc: Actor-critic reinforcement learning for nonlinear model predictive control. arXiv preprint arXiv:2406.03995, 2024

  23. [31]

    A reduction of imitation learning and structured prediction to no-regret online learning

    Stephane Ross, Geoffrey Gordon, and Drew Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. In Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, volume 15, pages 627--635. PMLR, 11-...

  24. [32]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Jascha Sohl-Dickstein, Eric Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep unsupervised learning using nonequilibrium thermodynamics. In Proceedings of the 32nd International Conference on Machine Learning, volume 37, pages 2256--2265. PMLR, 07--09 Jul 2015

  25. [33]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502, 2020

  26. [34]

    Efficient deep learning of robust policies from mpc using imitation and tube-guided data augmentation

    Andrea Tagliabue and Jonathan P How. Efficient deep learning of robust policies from mpc using imitation and tube-guided data augmentation. IEEE Transactions on Robotics, 2024

  27. [35]

    Deep generative models in robotics: A survey on learning from multimodal demonstrations

    Julen Urain, Ajay Mandlekar, Yilun Du, Mahi Shafiullah, Danfei Xu, Katerina Fragkiadaki, Georgia Chalvatzaki, and Jan Peters. Deep generative models in robotics: A survey on learning from multimodal demonstrations. arXiv preprint arXiv:2408.04380, 2024

  28. [36]

    Fatrop: A fast constrained optimal control problem solver for robot trajectory optimization and control

    Lander Vanroye, Ajay Sathya, Joris De Schutter, and Wilm Decr \'e . Fatrop: A fast constrained optimal control problem solver for robot trajectory optimization and control. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 10036--10043. ...

  29. [37]

    acados -- a modular open-source framework for fast embedded optimal control

    Robin Verschueren, Gianluca Frison, Dimitris Kouzoupis, Jonathan Frey, Niels van Duijkeren, Andrea Zanelli, Branimir Novoselnik, Thivaharan Albin, Rien Quirynen, and Moritz Diehl. acados -- a modular open-source framework for fast embedded optimal control. Mathematical Program...

  30. [38]

    Exploring model-based planning with policy networks

    Tingwu Wang and Jimmy Ba. Exploring model-based planning with policy networks. International Conference on Learning Representations, 2020

  31. [39]

    Wächter and L

    A. Wächter and L. Biegler. On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming. Math. Program, 106: 0 25--27, 2006

  32. [40]

    Fast nonlinear model predictive control on fpga using particle swarm optimization

    Fang Xu, Hong Chen, Xun Gong, and Qin Mei. Fast nonlinear model predictive control on fpga using particle swarm optimization. IEEE Transactions on Industrial Electronics, 63 0 (1): 0 310--321, 2016

  33. [41]

    Warm-start strategies in interior-point methods for linear programming

    E Alper Yildirim and Stephen J Wright. Warm-start strategies in interior-point methods for linear programming. SIAM Journal on Optimization, 12 0 (3): 0 782--810, 2002

  34. [42]

    Diffusion model predictive control

    Guangyao Zhou, Sivaramakrishnan Swaminathan, Rajkumar Vasudeva Raju, J Swaroop Guntupalli, Wolfgang Lehrach, Joseph Ortiz, Antoine Dedieu, Miguel L \'a zaro-Gredilla, and Kevin Murphy. Diffusion model predictive control. arXiv preprint arXiv:2410.05364, 2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.