Pith. sign in

REVIEW 3 major objections 5 minor 36 references

Bayesian Optimization Parameter Tuning Framework for a Lyapunov Based Path Following Controller

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper shows that Bayesian optimization, treating the vehicle as a black box, can tune four strongly coupled controller gains in 32 real-vehicle laps, improving the lap cost from 2076 to 1888 and cutting peak lateral error by about 18 pe

desk verdict Solid real-world BO tuning demo undermined by weak statistical evidence—single noisy laps, no repeats, and a stability proof gap. read the letter →

arxiv 2512.12649 v2 pith:ZLIOBN3E submitted 2025-12-14 cs.RO cs.SYeess.SY

Zhewen Zheng , Wenjing Cao , Hongkang Yu , Mo Chen , Takashi Suzuki This is my paper · ORCID
classification cs.ROcs.SYeess.SY
keywords BayesianoptimizationparametertuningGaussianprocessExpectedImprovementLyapunov-basedcontrolpathfollowingmobilerobotreal-worldexperiments
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

The paper sets out to show that Bayesian optimization is a practical way to tune the four coupled gains of a Lyapunov-based path-following controller on a real three-wheeled robot, where each evaluation costs one full lap. Using a Gaussian-process surrogate and expected-improvement acquisition, it claims to find gains within 32 trials (15 of them warm-start) that lower the lap-based tracking cost from 2076.35 to 1887.65 and reduce peak lateral error by around 18 percent. The result matters because manual tuning of such coupled nonlinear gains is slow and often blind, and simulation-only tuning does not transfer to hardware. If the claim holds, BO becomes a credible default tool for data-efficient controller tuning on real platforms.

What carries the argument

The load-bearing mechanism is the Bayesian optimization loop: a Gaussian-process surrogate with an ARD Matérn-5/2 kernel models the mapping from the four gain values to the lap cost, and expected improvement selects the next gain vector by balancing predicted gain against model uncertainty. Two engineering details make it work on hardware: gains are optimized in logarithmic space because their effective ranges differ by orders of magnitude, and any lap lost to instability is retained with a penalty proportional to the uncompleted distance, so the surrogate learns which regions are unsafe without discarding those trials.

What would settle it

Rerun the baseline gains and the Iteration-31 gains for several additional laps each under identical conditions and compare the distributions of lap costs; if the observed 1887.65 vs 2076.35 difference is smaller than the run-to-run spread, the reported improvement is not distinguishable from measurement noise.

Watch

Extended reading notes

Core claim

We demonstrate that Bayesian optimization, with a Gaussian-process surrogate, expected-improvement acquisition, logarithmic gain scaling, and a penalty for incomplete unstable laps, can tune the four coupled gains of a Lyapunov-based kinematic path-following controller on a real three-wheeled robot. Within a budget of 32 full-lap evaluations, the best BO-selected gains reach a performance cost of 1887.65 versus the manually tuned baseline of 2076.35, with the maximum lateral error reduced by about 18 percent and visibly smoother yaw-rate and speed profiles, especially on curved sections. This is presented as evidence that BO is a practical, data-efficient tool for black-box controller tuning

Load-bearing premise

A single full-lap evaluation of each gain setting reliably ranks gain vectors, even though the paper acknowledges that run-to-run disturbances can misrank vectors and that repeated trials cannot correct these misjudgements within the budget.

Editorial extensions

If this is right

  • If the central claim holds, engineers can tune nonlinear geometric controllers with a few dozen hardware experiments instead of hundreds of manual trials.
  • The same loop—GP surrogate, expected improvement, log-scale gains, penalty for failed runs—should transfer to other black-box closed-loop tuning tasks, such as MPC weights or PID gains on vehicles.
  • The warm-start and penalization design means unsafe gain regions can be mapped and avoided without discarding the information from unstable laps, reducing the risk of damaging hardware experiments.
  • The observed error reductions concentrate in curved track sections, indicating that BO-tuned gains improve transient tracking rather than only steady-state offset.
  • Because each trial is one lap, the approach is affordable on platforms where running a full experiment is costly but not destructive, opening the door to on-site retuning as conditions change.

Reading between the lines

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

  • The headline improvement (about 9% in cost) is measured on one lap per setting; given the paper's own acknowledgment of run-to-run disturbance, the practical gain may be smaller or even indistinguishable from noise until repeated trials confirm it.
  • The same framework could be extended to multi-objective costs, such as minimizing both tracking error and control effort, without changing the surrogate—suggesting a path to more general tuning policies.
  • A natural falsifier is to freeze the search after finding the best gain and rerun it multiple times to estimate the noise floor; if the noise floor exceeds the measured improvement, a different comparison protocol (e.g., paired laps or averaged costs) is needed before claiming superiority over manual tuning.
  • The paper's own future direction—safety-aware BO with a conservative, expanding safe region—could turn this tuning tool into an online adaptation layer that re-tunes gains as track or platform conditions drift.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 Bayesian optimization (BO) framework for tuning the four gains of a Lyapunov-based path-following controller on Honda's AI-Formula three-wheeled robot. The controller gains are treated as black-box inputs; a Gaussian-process surrogate with an ARD Matérn-5/2 kernel and expected-improvement acquisition is used to select gain vectors within a bounded search domain. After 15 warm-start evaluations (including a manually tuned baseline) and 17 acquisition-driven evaluations, the best observed performance is reported at Iteration 31 with cost 1887.65 versus the baseline cost 2076.35, and the paper claims that BO efficiently locates well-performing gain configurations with a small real-world evaluation budget.

Significance. If the empirical claim is supported, the paper would be a useful real-vehicle demonstration of BO tuning for a geometric/nonlinear path-following controller, extending earlier BO studies that focus mainly on MPC formulations. The paper has clear strengths: the BO machinery is standard and correctly described, the Lyapunov-derivative correction is explicit, the penalty formulation for unstable laps is a reasonable practical device, and the authors state that code is available. However, the central validation is currently weak. The reported improvement rests on single-run evaluations without repeated trials, error bars, or a random-search control, and the cost function includes a per-lap normalization that can itself distort comparisons. These issues make the headline contribution—that BO is reliably data-efficient for this controller—not yet established.

major comments (3)
  1. [Section IV-B, Fig. 5, Table II] The headline improvement is not supported by the experimental design because every gain vector is evaluated in exactly one full-lap run. The best cost (1887.65 at Iteration 31) is the minimum of 32 noisy one-lap evaluations, while the baseline (2076.35) is a single draw. Even if all settings have identical expected cost, the minimum of 32 noisy draws is expected to lie below a single draw, so part or all of the reported 9% gap may be a selection artifact. The manuscript itself concedes in Section IV-B that disturbances 'can misrank gain vectors' and that with limited budget 'these misjudgements cannot be corrected through repeated trials.' The authors should provide repeated laps at least at the baseline and at the final candidate (e.g., 3–5 laps each), report mean and standard deviation, and quantify the selection effect via a permutation test or a random-search comparison.
  2. [Section IV-B, Iterations 1–15 vs 16–32] The warm-start initialization already appears to find near-optimal settings: Iteration 9 yields cost 1985.14, only about 5% above the final best, and Iteration 7 yields 2019.88. Since the initial 15 evaluations are described as essentially a hybrid space-filling/randomized search, it is unclear whether the acquisition-driven BO loop adds value over the initial sample. The paper does not provide a control of 32 space-filling or random evaluations, nor does it report the full set of 32 points and costs so that the later iterations can be assessed independently. The conclusion that BO 'identifies well-performing configurations with far fewer trials' would require evidence that the GP/EI-guided iterations improve on the warm-start phase beyond random or space-filling selection.
  3. [Section II-D, Eqs. (21)–(23)] The cost normalization in Eqs. (22)–(23) is problematic. Because each lap's error series is divided by its own median, J_lat is approximately N * mean(|e_lat|) / median(|e_lat|). A lap with large absolute errors has a large denominator and is thereby penalized less, while a lap with uniformly small errors has a small denominator and is penalized more. This per-lap normalization makes J incomparable across gain vectors and can bias the GP toward regions with larger tracking errors. The authors should justify this normalization or replace it with a fixed reference normalization (e.g., using the baseline lap's median or absolute units with a known scale factor). As written, the cost values in Fig. 5 and Table II may not faithfully rank the tested gains.
minor comments (5)
  1. [Abstract and Section IV] The abstract says the framework is 'assessed through repeated full-lap experiments,' but every gain setting is evaluated only once. Rephrase to avoid implying replicated trials.
  2. [Section III-D] The hybrid space-filling strategy for Iterations 2–15 is described only verbally. Please provide the exact sampling method, bounds, and any random seed so the warm-start set is reproducible.
  3. [Section III-B] The GP hyperparameters (lengthscales, signal variance, noise variance) are said to be obtained by marginal-likelihood maximization but their fitted values are not reported. Given that each θ is observed once, σ_n² is not identifiable without further structure; reporting fitted values or fixing σ_n from separate experiments would improve reproducibility.
  4. [Section II-C, Eq. (18)] The angular velocity command contains sinα in the denominator, which is singular at α=0. The implementation must handle this division; please state the safeguard used on the robot (e.g., a small-angle approximation, a lower clamp, or a modified formula).
  5. [Fig. 5] The upper and lower subplots have the same x-label but the lower plot's y-range is not explicitly described in the caption. Add a caption note that the lower panel is a zoomed view and mark the baseline, Iteration 9, and Iteration 31 for clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported optimum is a measured hardware outcome, not a value produced by the GP or acquisition function; the sole self-citation is not load-bearing.

full rationale

The paper's central result is the experimentally evaluated cost bJBO = 1887.65 at Iteration 31, measured on the AI-Formula robot according to Eq. (39), compared with the measured baseline 2076.35. Algorithm 1 outputs the minimum of observed bJBO values (line 10); the GP posterior (Eqs. 32-34) and EI (Eq. 38) only propose which gain vector to try next. Hence the 'prediction' of good performance is not derived from the fitted surrogate by construction; it is a direct observation. The paper explicitly admits in Sec. IV-B that disturbances 'can misrank gain vectors' and that 'these misjudgements cannot be corrected through repeated trials' — a statistical limitation about single-lap, noisy evaluation, not a circularity. The only self-citation, [32], is used for the manual baseline and prior controller validation, but the baseline setting is re-evaluated in this paper with Eq. (39), and [32] does not enter the BO derivation or the final optimum. Thus there is no equation-level or citation-level circularity; concerns about best-of-32 selection effects are correctness/evidence issues, not circularity.

Assumptions & free parameters 6 free parameters · 7 assumptions · 0 invented entities

The central claim rests on the validity of the kinematic approximation, the Gaussian-noise GP model, the assumed stability of the underlying controller, and several hand-chosen experimental/objective parameters (w, lambda_pen, bounds, warm-start design). No new physical entities are introduced.

free parameters (6)
  • Cost normalization weight w = 0.1
    Hand-chosen in Eq. (21) to balance lateral and heading error terms in the objective.
  • Instability penalty lambda_pen = 7000
    Hand-chosen finite penalty in Eq. (39) for incomplete laps; affects how BO treats unsafe regions.
  • Search bounds theta_lb, theta_ub = lambda_v [1e-4,0.5], lambda_a [1e-3,1.5], k1 [1e-2,10], k2 [1e-1,100]
    User-specified admissible range for the four gains; shapes the feasible domain in Eq. (24).
  • Initial sample size Ninit and budget Nmax = 15 and 32
    Hand-chosen experiment budget; Ninit=15 includes one manual baseline plus 14 space-filling points.
  • GP hyperparameters = Not reported numerically
    Length scales, signal variance, and noise variance are fit by marginal likelihood in Eq. (35).
  • Hybrid space-filling design split = Iterations 2-7 near baseline; iterations 8-15 broad
    Hand-specified mixture of local refinement and global coverage for the warm-start set.
assumptions (7)
  • domain assumption The AI-Formula robot is accurately modeled as a non-holonomic two-wheel differential-drive mobile robot (Eqs. 1-10).
    Section II-B simplifies the front-wheel-drive, rear-caster platform to a unicycle model; real-wheel slip and caster effects are ignored.
  • domain assumption Measurement noise on the performance index is additive Gaussian with zero mean: epsilon_i ~ N(0, sigma_n^2).
    Eq. (27) assumes Gaussian noise for the GP observation model, though the text also describes non-Gaussian disturbances like passive rear-wheel heading.
  • domain assumption A zero-mean GP with ARD Matérn-5/2 kernel is an adequate surrogate for the cost function over the 4D log-gain space.
    Section III-B assigns this prior without a justification beyond standard practice; the cost surface is unknown.
  • domain assumption The Lyapunov-based controller is stable in practice for the tested gain region.
    Section II-C claims convergence, but the combined Lyapunov derivative has an unaddressed positive cross term; the claim is not fully proved. The paper relies on empirical validation from a self-cited preprint [32].
  • domain assumption Variations in measured performance are mainly due to controller gains because the track, speed, and sensor configuration are fixed.
    Section IV-A asserts repeatability, yet later states that run-to-run disturbances are significant enough to misrank gain vectors.
  • ad hoc to paper Unstable laps can be meaningfully incorporated into the GP by adding the finite penalty lambda_pen.
    Eq. (39) encodes instability as a penalized cost; the choice of penalty value and its effect on the GP fit are not analyzed.
  • ad hoc to paper Normalizing each lap's errors by their own median produces a comparable objective across laps.
    Eqs. (22)-(23) use per-lap medians, which changes the objective scale from lap to lap and could reduce comparability.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bayesian Optimization Parameter Tuning Framework for a Lyapunov Based Path Following Controller." pith.science (2026). https://pith.science/paper/ZLIOBN3E

@misc{pith2026251212649,
  author       = {Pith},
  title        = {Pith review of: Bayesian Optimization Parameter Tuning Framework for a Lyapunov Based Path Following Controller},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZLIOBN3E}},
  note         = {Machine review of arXiv:2512.12649}
}
read the original abstract

Parameter tuning in real-world experiments is constrained by the limited evaluation budget available on hardware. The path-following controller studied in this paper reflects a typical situation in nonlinear geometric controller, where multiple gains influence the dynamics through coupled nonlinear terms. Such interdependence makes manual tuning inefficient and unlikely to yield satisfactory performance within a practical number of trials. To address this challenge, we propose a Bayesian optimization (BO) framework that treats the closed-loop system as a black box and selects controller gains using a Gaussian-process surrogate. BO offers model-free exploration, quantified uncertainty, and data-efficient search, making it well suited for tuning tasks where each evaluation is costly. The framework is implemented on Honda's AI-Formula three-wheeled robot and assessed through repeated full-lap experiments on a fixed test track. The results show that BO improves controller performance within 32 trials, including 15 warm-start initial evaluations, indicating that it can efficiently locate high-performing regions of the parameter space under real-world conditions. These findings demonstrate that BO provides a practical, reliable, and data-efficient tuning approach for nonlinear path-following controllers on real robotic platforms.

Figures

Figures reproduced from arXiv: 2512.12649 by the authors.

Figure 1
Figure 1. Overall view and hardware configuration of the AI-Formula robot platform [30]. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Non-holonomic two-wheel differential-drive mobile robot model. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overview of the proposed BO framework for path following controller tuning. The acquisition function selects a parameter vector [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Overview of the real-world test course. BO to account for the different scales of the gains and prevents the loss of resolution inherent in a linear scaling. Safety handling and penalization strategy: A small fraction of gain combinations inevitably produce unstable cl…
Figure 5
Figure 5. Figure 5: Experimentally evaluated performance JbBO(θ) over iterations [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Path comparison across selected BO iterations. The plot shows the [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: State trajectories for Iterations 1, 9 and 31 along the normalized path progress. The left column shows the full-lap profiles of velocity, yaw rate, [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 6
Figure 6. Figure 6: In the velocity plots, the manually tuned baseline [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references

  1. [1]

    N. A. I. Ruslan, N. H. Amer, K. Hudha, Z. Abdul Kadir, S. A. F. Mohamed Ishak, and S. M. F. Syed Dardin. ”Modelling and control strategies in path tracking control for autonomous tracked vehicles: A review of state of the art and challenges,”Journal of Terramechanics, vol. 105, pp. 67–79, 2023

    2023

  2. [2]

    A review of some pure- pursuit based path tracking techniques for control of autonomous vehicle,

    M. Samuel, M. Hussein, and M. B. Mohamad. “A review of some pure- pursuit based path tracking techniques for control of autonomous vehicle,”

  3. [3]

    A review of PID control, tuning methods and applications,

    R. P. Borase, D. K. Maghade, S. Y . Sondkar, et al. “A review of PID control, tuning methods and applications,”International Journal of Dynamics and Control, vol. 9, no. 2, pp. 818–827, 2021

    2021

  4. [4]

    Review of intelligent motor controller parameter self-tuning technology,

    Z. Song and Y . Huang. “Review of intelligent motor controller parameter self-tuning technology,”Electronics, vol. 14, no. 11, p. 2229, 2025

    2025

  5. [5]

    Review and perfor- mance evaluation of path tracking controllers of autonomous vehicles,

    M. Rokonuzzaman, N. Mohajer, S. Nahavandi, et al. “Review and perfor- mance evaluation of path tracking controllers of autonomous vehicles,” IET Intelligent Transport Systems, vol. 15, no. 5, pp. 646–670, 2021

    2021

  6. [6]

    Trajectory tracking control of multirotors from modelling to experiments: A survey,

    H. Lee and H. J. Kim. “Trajectory tracking control of multirotors from modelling to experiments: A survey,”International Journal of Control, Automation and Systems, vol. 15, no. 1, pp. 281–292, 2017

    2017

  7. [7]

    A review of adaptive tuning of PID-controller: Optimization techniques and applications,

    I. A. Abbas and M. K. Mustafa. “A review of adaptive tuning of PID-controller: Optimization techniques and applications,”International Journal of Nonlinear Analysis and Applications, vol. 15, no. 2, pp. 29–37, 2024

    2024

  8. [8]

    ROS-based trajectory tracking control for autonomous tracked vehicle using opti- mized backstepping and sliding mode control,

    A. D. Sabiha, M. A. Kamel, E. Said, and W. M. Hussein. “ROS-based trajectory tracking control for autonomous tracked vehicle using opti- mized backstepping and sliding mode control,”Robotics and Autonomous Systems, vol. 152, p. 104058, 2022

    2022

Show all 36 references
  1. [9]

    Lyapunov-based approach for target track- ing control of a mobile robot,

    F. M. Kasim and V . R. Jisha. “Lyapunov-based approach for target track- ing control of a mobile robot,” in2015 IEEE International Conference on Electrical, Computer and Communication Technologies (ICECCT), Coimbatore, India, pp. 1–6, 2015

    2015

  2. [10]

    Path tracking and direct yaw moment coordinated control based on robust MPC with the finite time horizon for autonomous independent-drive vehicles,

    H. Peng, W. Wang, Q. An, C. Xiang, and L. Li. “Path tracking and direct yaw moment coordinated control based on robust MPC with the finite time horizon for autonomous independent-drive vehicles,”IEEE Transactions on Vehicular Technology, vol. 69, no. 6, pp. 6053–6066, 2020

    2020

  3. [11]

    USV formation and path-following control via deep reinforcement learning with random braking,

    Y . Zhao, Y . Ma, and S. Hu. “USV formation and path-following control via deep reinforcement learning with random braking,”IEEE Transactions on Neural Networks and Learning Systems, vol. 32, no. 12, pp. 5468– 5478, 2021

    2021

  4. [12]

    Modelling and Control Strategies in Path Tracking Control for Autonomous Ground Vehicles: A Review of State of the Art and Challenges,

    N. H. Amer, H. Zamzuri, K. Hudha, et al. “Modelling and Control Strategies in Path Tracking Control for Autonomous Ground Vehicles: A Review of State of the Art and Challenges,”J Intell Robot Syst, vol. 86, pp. 225–254, 2017

    2017

  5. [13]

    Iterative feedback tuning—an overview,

    H. Hjalmarsson. “Iterative feedback tuning—an overview,”International Journal of Adaptive Control and Signal Processing, vol. 16, no. 5, pp. 373–395, 2002

    2002

  6. [14]

    Gradient-descent-based learning gain for backstepping controller and disturbance observer of nonlinear systems,

    S. You, Y . S. Son, Y . Gui, and W. Kim. “Gradient-descent-based learning gain for backstepping controller and disturbance observer of nonlinear systems,”IEEE Access, vol. 11, pp. 2743–2753, 2023

    2023

  7. [15]

    DiffLoop: Tuning PID controllers by differentiating through the feedback loop,

    A. R. Kumar and P. J. Ramadge. “DiffLoop: Tuning PID controllers by differentiating through the feedback loop,” in2021 55th Annual Conference on Information Sciences and Systems (CISS), Baltimore, MD, USA, pp. 1–6, 2021

    2021

  8. [16]

    Evolutionary algorithms in control systems engineering: a survey,

    P. J. Fleming and R. C. Purshouse. “Evolutionary algorithms in control systems engineering: a survey,”Control Engineering Practice, vol. 10, no. 11, pp. 1223–1241, 2002

    2002

  9. [17]

    Tuning of PID controller based on a multiobjective genetic algorithm applied to a robotic manipu- lator,

    H. V . H. Ayala and L. dos Santos Coelho. “Tuning of PID controller based on a multiobjective genetic algorithm applied to a robotic manipu- lator,”Expert Systems with Applications, vol. 39, no. 10, pp. 8968–8974, 2012

    2012

  10. [18]

    Tuning PID and PIλDδ controllers using the integral time absolute error criterion,

    D. Maiti, A. Acharya, M. Chakraborty, A. Konar, and R. Janarthanan. “Tuning PID and PIλDδ controllers using the integral time absolute error criterion,” inInternational Conference on Information and Automation, pp. 457–462, 2008

    2008

  11. [19]

    Neumann-Brosig, A

    M. Neumann-Brosig, A. Marco, D. Schwarzmann, et al. ‘Data-efficient autotuning with Bayesian optimization: An industrial control study,’IEEE Transactions on Control Systems Technology, vol. 28, no. 3, pp. 730–740, 2019

    2019

  12. [20]

    J. P. L. Coutinho, L. O. Santos, and M. S. Reis. ‘Bayesian Optimization for automatic tuning of digital multi-loop PID controllers,’Computers & Chemical Engineering, vol. 173, p. 108211, 2023

    2023

  13. [21]

    Shahriari, K

    B. Shahriari, K. Swersky, Z. Wang, et al. ‘Taking the human out of the loop: A review of Bayesian optimization,’Proceedings of the IEEE, vol. 104, no. 1, pp. 148–175, 2015. JOURNAL OF LATEX CLASS FILES, VOL. 18, NO. 9, SEPTEMBER 2020 12

    2015

  14. [22]

    Greenhill, S

    S. Greenhill, S. Rana, S. Gupta, et al. ‘Bayesian optimization for adaptive experimental design: A review,’IEEE Access, vol. 8, pp. 13937– 13948, 2020

    2020

  15. [23]

    Tesch, J

    M. Tesch, J. Schneider, and H. Choset. ‘Using response surfaces and expected improvement to optimize snake robot gait parameters,’ inPro- ceedings of the 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, pp. 1069–1074, 2011

    2011

  16. [24]

    Calandra, N

    R. Calandra, N. Gopalan, A. Seyfarth, et al. ‘Bayesian gait optimization for bipedal locomotion,’ inProceedings of the International Conference on Learning and Intelligent Optimization, Cham: Springer International Publishing, pp. 274–290, 2014

    2014

  17. [25]

    Schillinger, B

    M. Schillinger, B. Hartmann, P. Skalecki, et al. ‘Safe active learn- ing and safe Bayesian optimization for tuning a PI-controller,’IFAC- PapersOnLine, vol. 50, no. 1, pp. 5967–5972, 2017

    2017

  18. [26]

    Gharib, D

    A. Gharib, D. Stenger, R. Ritschel, et al. ‘Multi-objective optimization of a path-following MPC for vehicle guidance: A Bayesian optimization approach,’ inProceedings of the 2021 European Control Conference (ECC), IEEE, pp. 2197–2204, 2021

    2021

  19. [27]

    Str ´o˙zecki, N

    W. Str ´o˙zecki, N. Ait Oufroukh, Y . Kebbati, et al. ‘Automatic tuning of MPC for autonomous vehicle using Bayesian optimization,’ inProceed- ings of the 2021 IEEE International Conference on Networking, Sensing and Control (ICNSC), IEEE, vol. 1, pp. 1–6, 2021

    2021

  20. [28]

    L. P. Fr ¨ohlich, C. K ¨uttel, E. Arcari, et al. ‘Contextual tuning of model predictive control for autonomous racing,’ inProceedings of the 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, pp. 10555–10562, 2022

    2022

  21. [29]

    J. Menn, P. Pelizzari, M. Fleps-Dezasse, et al. ‘Lipschitz safe Bayesian optimization for automotive control,’ inProceedings of the 2024 IEEE 63rd Conference on Decision and Control (CDC), IEEE, pp. 3782–3788, 2024

    2024

  22. [30]

    Okada, Y

    M. Okada, Y . Akimoto, A. Kato, and Y . Yasui. ‘AI Formula,’ in Proceedings of the 2024 Spring Conference of the Society of Automotive Engineers of Japan (JSAE) – Academic Lectures, no. 20245167, 2024

    2024

  23. [31]

    Veronesi and A

    M. Veronesi and A. Visioli. ‘Performance assessment and retuning of PID controllers for integral processes,’Journal of Process Control, vol. 20, no. 3, pp. 261–269, 2010

    2010

  24. [32]

    M. Chen. ‘Integrated YOLOP perception and Lyapunov-based control for autonomous mobile robot navigation on track,’ arXiv:2512.01608, 2025

    2025

  25. [33]

    Bliek, A

    L. Bliek, A. Guijt, R. Karlsson, et al. ‘Benchmarking surrogate-based optimisation algorithms on expensive black-box functions,’Applied Soft Computing, vol. 147, p. 110744, 2023

    2023

  26. [34]

    Marrel and B

    A. Marrel and B. Iooss. ‘Probabilistic surrogate modeling by Gaussian process: A review on recent insights in estimation and validation,’ Reliability Engineering & System Safety, vol. 247, p. 110094, 2024

    2024

  27. [35]

    Snoek, H

    J. Snoek, H. Larochelle, and R. P. Adams. ‘Practical Bayesian optimiza- tion of machine learning algorithms,’Advances in Neural Information Processing Systems, vol. 25, 2012

    2012

  28. [36]

    C. E. Rasmussen. ‘Gaussian processes in machine learning,’ inSum- mer School on Machine Learning, Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 63–71, 2003. Zhewen ZhengZhewen Zheng received the B.Eng. degree in mechanical design, manufacture, and au- tomation from the U...

    2003

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.