Pith. sign in

REVIEW 3 major objections 3 minor 42 references

CoCoNav: Conformal Control for Safe Robot Navigation in Crowds

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

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

Pith's one-line read CoCoNav separates planning from safety certification to keep robot navigation safe and feasible.

desk verdict CoCoNav is a promising, novel combination of online conformal calibration and relax-then-verify MPC that deserves referee time, but the safety certificate's coverage-under-shift assumption is the thing to scrutinize. read the letter →

arxiv 2608.07751 v1 pith:CZOYS2JC submitted 2026-08-07 cs.RO cs.LG

Cheng Guo , Mingzhe Ni , Zheng Liang , Yihu Ling , Yuan Hu , Michele Caprio , Daniele Pucci , Wei Pan This is my paper · ORCID
classification cs.ROcs.LG
keywords crowdnavigationconformalpredictionmodelpredictivecontrolsafetycertificatedistributionshiftuncertaintyquantificationquadrupedrobot
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

This paper proposes a crowd-navigation framework, CoCoNav, that aims to make robot navigation safe in crowded spaces without sacrificing feasibility or efficiency. It claims that by separating the planning problem from the safety-certification problem, a robot can always generate a nominal trajectory because soft-constrained MPC is always feasible, while a separate verification step certifies that trajectory and its contingency maneuvers against statistically calibrated prediction-error bounds. The calibration is performed online by a horizon-specific conformal proportional-integral controller that adjusts error bounds so their long-run empirical coverage stays at a target level. Simulations and quadruped experiments are presented as evidence that this approach achieves a favorable balance among collision avoidance, task success, and navigation efficiency.

What carries the argument

The central mechanism is a two-stage relax-then-verify planning loop. Stage one uses soft-constrained MPC to generate nominal trajectories without hard safety constraints, so the solver always has a feasible solution. Stage two certifies the nominal trajectory and contingency maneuvers against calibrated trajectory-error bounds before execution. The bounds come from a horizon-specific conformal proportional-integral controller, which is a feedback loop that treats the empirical coverage of prediction errors as the controlled variable and adjusts quantile-based conformal bounds for each horizon to track a target coverage level. This decouples the planner's feasibility problem from the statistical validity problem, allowing safety certification to remain meaningful even as pedestrian behavior changes.

What would settle it

Deploy CoCoNav in a crowd and then introduce a sudden, persistent change in pedestrian motion, for example all pedestrians abruptly adopt a new avoidance rule or double their walking speed. Record the empirical coverage of the trajectory-error bounds over a long horizon: if the coverage rate drops substantially below the target and stays there, or if a collision occurs that the certificate should have ruled out, then the coverage-regulation claim is falsified.

Watch

Extended reading notes

Core claim

The central claim is that a robot can navigate through crowds safely even when pedestrian-motion predictions are unreliable, by making the safety guarantee a separate statistical object rather than a hard constraint in the optimizer. The planner uses soft-constrained MPC to generate nominal trajectories, so the optimization remains feasible; before execution, a verifier checks the nominal trajectory and its contingency maneuvers against error bounds produced by an online conformal PI controller. The controller dynamically widens or narrows the bounds per horizon to keep the empirical coverage of trajectory predictions at the desired long-run rate, which lets the framework respond to shifting prediction errors. The claimed result is a favorable balance among collision avoidance, task success, and navigation efficiency in simulations and quadruped experiments.

Load-bearing premise

The load-bearing premise is that the online conformal coverage controller maintains its long-run coverage guarantee even when pedestrian behavior shifts, which requires unstated assumptions about how quickly or smoothly the prediction-error distribution can change.

Editorial extensions

If this is right

  • Robot navigation can remain safe even when the pedestrian prediction model is misspecified, because the error bounds are calibrated to observed data rather than assumed in advance.
  • MPC solver feasibility no longer competes with safety: because constraints are soft during planning and safety is verified separately, a trajectory can always be found even in dense crowds.
  • Long-run coverage regulation lets the framework adapt to shifting prediction errors, avoiding both over-conservative bounds that make navigation sluggish and over-confident bounds that risk collisions.
  • The relax-then-verify design can be reused in other safety-critical planning problems where hard constraint tightening would make the optimization infeasible.

Reading between the lines

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

  • This framework suggests a general template: first generate candidate trajectories with an unconstrained or soft-constrained optimizer, then filter them through a statistical safety certificate before execution, which could apply well beyond crowd navigation.
  • A natural testable extension is to apply the horizon-specific conformal PI controller to other prediction tasks, such as human motion forecasting in shared autonomy, and check whether the coverage regulation property persists.
  • Separating planning from certification opens the door to using planners that are not differentiable or not based on MPC, since the certificate is agnostic to how the candidate trajectory was produced.
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 / 3 minor

Summary. This abstract-only review assesses CoCoNav, a crowd-navigation framework that pairs a horizon-specific conformal proportional--integral controller (which adapts trajectory-error bounds to regulate long-run empirical coverage) with a relax-then-verify planner: soft-constrained MPC generates nominal trajectories, and a separate verification step certifies these trajectories, together with contingency maneuvers, against the calibrated bounds. The authors claim that this preserves MPC feasibility while providing runtime-certified safety, and report favorable results in simulations and quadruped experiments. Because the full text is not available, the assessment is limited to the claims and definitions presented in the abstract.

Significance. If the central claims are substantiated, the paper addresses a real and recognized problem: balancing collision avoidance, task success, and navigation efficiency under uncertain and shifting pedestrian behavior without sacrificing MPC feasibility. The separation of planning and certification is an elegant structural idea, and the use of online conformal calibration to avoid restrictive error models is a promising direction. The inclusion of both simulation and quadruped experiments is also a strength in principle. However, the abstract alone does not provide enough evidence to assess whether the safety certificate is mathematically valid or whether the experimental claims are statistically supported.

major comments (3)
  1. [Abstract] The safety certificate's validity rests on the conformal PI controller's 'long-run empirical coverage' of trajectory prediction errors under shifting pedestrian behavior. The abstract asserts this property but states no assumptions on the pattern or rate of distribution shift; without assumptions such as bounded total variation or a finite-horizon error bound, a burst of undercoverage can occur and the certificate may certify genuinely unsafe trajectories. The full text must state the exact coverage guarantee and its conditions.
  2. [Abstract] The claim that 'Simulations and quadruped experiments show that CoCoNav achieves a favorable balance' is not testable from the abstract, which reports no quantitative metrics, error bars, or statistical comparisons against baselines. A journal submission should report numeric collision rates, success rates, navigation times, and confidence intervals, together with baseline comparisons, to support this claim.
  3. [Abstract] The phrase 'certifying them, together with contingency maneuvers, against the calibrated bounds' is underspecified: the certificate is meaningful only if the contingency maneuvers are dynamically feasible within the robot's actuation limits and within the relevant time horizon, and if the verified set is a true reachability or safety guarantee. The abstract does not state what the verification step actually computes (e.g., forward reachable sets, invariant sets, or braking distances), so the safety-preservation property of 'relax-then-verify' is not yet established.
minor comments (3)
  1. [Abstract] The term 'long-run empirical coverage' should be defined precisely (e.g., coverage over a moving window, the length of the window, and whether it is an average over time or over all pedestrians).
  2. [Abstract] The phrase 'horizon-specific conformal proportional--integral controller' is dense; a brief explanation of how the PI gains affect the coverage adaptation would improve readability.
  3. [Abstract] The expression 'relax-then-verify' is suggestive but not self-explanatory; the abstract should clarify the relationship between the soft-constrained nominal trajectory and the certified trajectory (e.g., whether the verified trajectory is a modification of the nominal one).

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: conformal calibration and trajectory certification are distinct statistical and geometric steps.

full rationale

The abstract describes a two-stage pipeline: a horizon-specific conformal proportional-integral controller calibrates trajectory-error bounds to regulate long-run empirical coverage, and a relax-then-verify planner first generates nominal trajectories with soft-constrained MPC and then separately certifies them, together with contingency maneuvers, against those calibrated bounds. There is no equation or definition in the abstract that makes the certified safety property identical to the fitted bounds, nor is any parameter fitted to the collision outcomes that the certificate is meant to predict. The conformal calibration step is a standard statistical procedure whose input is prediction error and whose output is an uncertainty interval; using that interval as a verification constraint is a legitimate composition of statistical inference and geometric checking, not a circular reduction. Even if the long-run coverage guarantee requires unstated assumptions about distribution shift, that is a question of whether the conformal guarantee holds in the intended operating conditions, not whether the derivation reduces to its own inputs. No load-bearing self-citation appears in the abstract. Given the abstract-only evidence, there is no specific reduction to quote, and the appropriate finding is no significant circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central claims rest on standard robotic control assumptions (exchangeability or adaptivity of prediction errors, availability of state estimates, real-time verification) and on a small set of tunable parameters (coverage target, PI gains, MPC weights) whose values are not given. An abstract-only review cannot audit them further.

free parameters (3)
  • conformal coverage target (alpha) = not stated
    The conformal PI controller regulates long-run empirical coverage, which implies a chosen target coverage level (e.g., 0.90). Such levels are typically selected by the practitioner and affect the bound width.
  • conformal PI controller gains = not stated
    The horizon-specific conformal PI controller adjusts trajectory-error bounds; its proportional and integral gains are design parameters that must be chosen (likely tuned on simulation or validation data).
  • MPC horizon and soft-constraint weights = not stated
    The relax-then-verify planner uses soft-constrained MPC; the horizon length and penalty weights for soft constraints and tracking objective are user-chosen parameters that influence feasibility and behavior.
assumptions (3)
  • domain assumption Pedestrian prediction errors are exchangeable, or the adaptive conformal procedure maintains valid coverage under distribution shift.
    The conformal PI controller aims to regulate long-run empirical coverage; conformal guarantees classically rely on exchangeability, and online adaptations make additional assumptions (e.g., bounded shift rate). The abstract does not state these assumptions.
  • domain assumption The perception and tracking pipeline provides accurate enough state estimates for the robot and pedestrians to define prediction errors.
    Conformal calibration requires measuring trajectory prediction errors, which requires a state estimation pipeline. The abstract does not discuss sensor noise or tracking uncertainties.
  • ad hoc to paper The verification procedure can certify safety against the calibrated bounds in real time and the robot's actuation limits permit the computed contingency maneuvers.
    The relax-then-verify planner assumes the certifier is computationally feasible and contingency maneuvers are dynamically feasible, which is not demonstrated in the abstract.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CoCoNav: Conformal Control for Safe Robot Navigation in Crowds." pith.science (2026). https://pith.science/paper/CZOYS2JC

@misc{pith2026260807751,
  author       = {Pith},
  title        = {Pith review of: CoCoNav: Conformal Control for Safe Robot Navigation in Crowds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CZOYS2JC}},
  note         = {Machine review of arXiv:2608.07751}
}
read the original abstract

Safe and efficient robot navigation in crowds requires anticipating pedestrian motion despite uncertain and potentially shifting prediction errors. Existing reactive methods can produce oscillatory behavior, while predictive planners often treat forecasts as exact or rely on restrictive error models. Incorporating conservative uncertainty sets as hard constraints can also render model predictive control (MPC) infeasible. We propose \textit{CoCoNav}, a crowd-navigation framework that combines online conformal calibration with runtime-certified planning. A horizon-specific conformal proportional--integral controller adapts trajectory-error bounds to regulate long-run empirical coverage, enabling the framework to respond to changing prediction errors. A \textit{relax-then-verify} planner preserves solver feasibility by generating nominal trajectories with soft-constrained MPC and separately certifying them, together with contingency maneuvers, against the calibrated bounds before execution. Simulations and quadruped experiments show that CoCoNav achieves a favorable balance among collision avoidance, task success, and navigation efficiency relative to the evaluated baselines.

Figures

Figures reproduced from arXiv: 2608.07751 by the authors.

Figure 1
Figure 1. Safe navigation around dynamic agents. ( [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of CoCoNav. Online conformal calibration constructs adaptive robot no-go regions for soft MPC planning. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Snapshots of the environment at different levels. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Sequential snapshots of robot behaviors produced by our approach and baselines when navigating through crowds. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Comparison of robot motion generation between our method (top) and ACP-MPC (bottom): trajectory (left), linear [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Ablation of conformal safety margins under distri [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 24 canonical work pages

  1. [1]

    Principles and guidelines for evaluating social robot navigation algorithms,

    A. Francis, C. P ´erez-d’Arpino, C. Li, F. Xia, A. Alahi, R. Alami, A. Bera, A. Biswas, J. Biswas, R. Chandraet al., “Principles and guidelines for evaluating social robot navigation algorithms,”ACM Transactions on Human-Robot Interaction, vol. 14, no. 2, pp. 1–65, 2025

    2025

  2. [2]

    Model predictive contouring control for collision avoidance in unstructured dynamic environments,

    B. Brito, B. Floor, L. Ferranti, and J. Alonso-Mora, “Model predictive contouring control for collision avoidance in unstructured dynamic environments,”IEEE Robotics and Automation Letters, vol. 4, no. 4, pp. 4459–4466, 2019

    2019

  3. [3]

    From crowd motion prediction to robot navigation in crowds,

    S. Poddar, C. Mavrogiannis, and S. S. Srinivasa, “From crowd motion prediction to robot navigation in crowds,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2023, pp. 6765–6772

    work page 2023

  4. [4]

    Confidence-aware motion prediction for real-time collision avoidance1,

    D. Fridovich-Keil, A. Bajcsy, J. F. Fisac, S. L. Herbert, S. Wang, A. D. Dragan, and C. J. Tomlin, “Confidence-aware motion prediction for real-time collision avoidance1,”The International Journal of Robotics Research, vol. 39, no. 2-3, pp. 250–265, 2020

    work page 2020

  5. [5]

    Stochastic mpc with multi-modal predictions for traffic intersections,

    S. H. Nair, V . Govindarajan, T. Lin, C. Meissen, H. E. Tseng, and F. Borrelli, “Stochastic mpc with multi-modal predictions for traffic intersections,” in2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC). IEEE, 2022, pp. 635–640

    work page 2022

  6. [6]

    Conformal prediction: A gentle introduction,

    A. N. Angelopoulos, S. Bateset al., “Conformal prediction: A gentle introduction,”Foundations and trends® in machine learning, vol. 16, no. 4, pp. 494–591, 2023

    work page 2023

  7. [7]

    Adaptive conformal inference under dis- tribution shift,

    I. Gibbs and E. Candes, “Adaptive conformal inference under dis- tribution shift,”Advances in Neural Information Processing Systems, vol. 34, pp. 1660–1672, 2021

    work page 2021

  8. [8]

    Conformal pid control for time series prediction,

    A. Angelopoulos, E. Candes, and R. J. Tibshirani, “Conformal pid control for time series prediction,”Advances in neural information processing systems, vol. 36, pp. 23 047–23 074, 2023

    work page 2023

Show all 42 references
  1. [9]

    The safety filter: A unified view of safety-critical control in autonomous systems,

    K.-C. Hsu, H. Hu, and J. F. Fisac, “The safety filter: A unified view of safety-critical control in autonomous systems,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 7, 2023

    2023

  2. [10]

    A predictive safety filter for learning-based control of constrained nonlinear dynamical systems,

    K. P. Wabersich and M. N. Zeilinger, “A predictive safety filter for learning-based control of constrained nonlinear dynamical systems,” Automatica, vol. 129, p. 109597, 2021

    2021

  3. [11]

    Conflict avoidance in social navigation—a survey,

    R. Mirsky, X. Xiao, J. Hart, and P. Stone, “Conflict avoidance in social navigation—a survey,”ACM Transactions on Human-Robot Interaction, vol. 13, no. 1, pp. 1–36, 2024

    2024

  4. [12]

    Social force model for pedestrian dynamics,

    D. Helbing and P. Molnar, “Social force model for pedestrian dynamics,” Physical review E, vol. 51, no. 5, p. 4282, 1995

    1995

  5. [13]

    The dynamic window approach to collision avoidance,

    D. Fox, W. Burgard, and S. Thrun, “The dynamic window approach to collision avoidance,”IEEE robotics & automation magazine, vol. 4, no. 1, pp. 23–33, 2002

    2002

  6. [14]

    Reciprocal n- body collision avoidance,

    J. Van Den Berg, S. J. Guy, M. Lin, and D. Manocha, “Reciprocal n- body collision avoidance,” inRobotics Research: The 14th International Symposium ISRR. Springer, 2011, pp. 3–19

    2011

  7. [15]

    Topology-driven parallel trajectory optimization in dynamic environ- ments,

    O. De Groot, L. Ferranti, D. M. Gavrila, and J. Alonso-Mora, “Topology-driven parallel trajectory optimization in dynamic environ- ments,”IEEE Transactions on Robotics, 2024

    2024

  8. [16]

    Shine: Social homology identification for navigation in crowded environments,

    D. Martinez-Baselga, O. de Groot, L. Knoedler, L. Riazuelo, J. Alonso- Mora, and L. Montano, “Shine: Social homology identification for navigation in crowded environments,”The International Journal of Robotics Research, p. 02783649251344639, 2024

    2024

  9. [17]

    Towards safe navigation through crowded dynamic environments,

    Z. Xie, P. Xin, and P. Dames, “Towards safe navigation through crowded dynamic environments,” in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2021, pp. 4934–4940

    2021

  10. [18]

    Deep local trajectory replanning and control for robot navigation,

    A. Pokle, R. Mart ´ın-Mart´ın, P. Goebel, V . Chow, H. M. Ewald, J. Yang, Z. Wang, A. Sadeghian, D. Sadigh, S. Savareseet al., “Deep local trajectory replanning and control for robot navigation,” in2019 international conference on robotics and automation (ICRA). IEEE, 2019, pp...

    2019

  11. [19]

    Socially compliant navigation dataset (scand): A large-scale dataset of demonstrations for social navigation,

    H. Karnan, A. Nair, X. Xiao, G. Warnell, S. Pirk, A. Toshev, J. Hart, J. Biswas, and P. Stone, “Socially compliant navigation dataset (scand): A large-scale dataset of demonstrations for social navigation,”IEEE Robotics and Automation Letters, vol. 7, no. 4, pp. 11 807–11 814, 2022

    2022

  12. [20]

    Towards imitation learning in real world unstructured social mini-games in pedestrian crowds,

    R. Chandra, H. Karnan, N. Mehr, P. Stone, and J. Biswas, “Towards imitation learning in real world unstructured social mini-games in pedestrian crowds,”arXiv e-prints, pp. arXiv–2405, 2024

    2024

  13. [21]

    Crowd-robot interaction: Crowd-aware robot navigation with attention-based deep reinforcement learning,

    C. Chen, Y . Liu, S. Kreiss, and A. Alahi, “Crowd-robot interaction: Crowd-aware robot navigation with attention-based deep reinforcement learning,” in2019 international conference on robotics and automation (ICRA). IEEE, 2019, pp. 6015–6022

    2019

  14. [22]

    Collision avoidance in pedestrian-rich environments with deep reinforcement learning,

    M. Everett, Y . F. Chen, and J. P. How, “Collision avoidance in pedestrian-rich environments with deep reinforcement learning,”Ieee Access, vol. 9, pp. 10 357–10 377, 2021

    2021

  15. [23]

    Social lstm: Human trajectory prediction in crowded spaces,

    A. Alahi, K. Goel, V . Ramanathan, A. Robicquet, L. Fei-Fei, and S. Savarese, “Social lstm: Human trajectory prediction in crowded spaces,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 961–971

    2016

  16. [24]

    Trajec- tron++: Dynamically-feasible trajectory forecasting with heterogeneous data,

    T. Salzmann, B. Ivanovic, P. Chakravarty, and M. Pavone, “Trajec- tron++: Dynamically-feasible trajectory forecasting with heterogeneous data,” inEuropean Conference on Computer Vision. Springer, 2020, pp. 683–700

    2020

  17. [25]

    Socialvae: Human trajec- tory prediction using timewise latents,

    P. Xu, J.-B. Hayet, and I. Karamouzas, “Socialvae: Human trajec- tory prediction using timewise latents,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 511–528

    2022

  18. [26]

    Step: Stochastic traversability evaluation and planning for risk-aware off-road navigation,

    D. D. Fan, K. Otsu, Y . Kubo, A. Dixit, J. Burdick, and A.-A. Agha- Mohammadi, “Step: Stochastic traversability evaluation and planning for risk-aware off-road navigation,”arXiv preprint arXiv:2103.02828, 2021

    2021 arXiv

  19. [27]

    Gaussian process-based visual pursuit control with unknown target motion learning in three dimensions,

    M. Omainska, J. Yamauchi, T. Beckers, T. Hatanaka, S. Hirche, and M. Fujita, “Gaussian process-based visual pursuit control with unknown target motion learning in three dimensions,”SICE Journal of Control, Measurement, and System Integration, vol. 14, no. 1, pp. 116–127, 2021

    2021

  20. [28]

    Online update of safety assurances using confidence-based predictions,

    K. Nakamura and S. Bansal, “Online update of safety assurances using confidence-based predictions,”arXiv preprint arXiv:2210.01199, 2022

    2022 arXiv

  21. [29]

    Safe planning in dynamic environments using conformal prediction,

    L. Lindemann, M. Cleaveland, G. Shim, and G. J. Pappas, “Safe planning in dynamic environments using conformal prediction,”IEEE Robotics and Automation Letters, vol. 8, no. 8, pp. 5116–5123, 2023

    2023

  22. [30]

    Interaction-aware conformal prediction for crowd navigation,

    Z. Huang, T. Ji, H. Zhang, F. C. Pouria, K. Driggs-Campbell, and R. Dong, “Interaction-aware conformal prediction for crowd navigation,” arXiv preprint arXiv:2502.06221, 2025

    2025 arXiv

  23. [31]

    Conformal predictive safety filter for rl controllers in dynamic environments,

    K. J. Strawn, N. Ayanian, and L. Lindemann, “Conformal predictive safety filter for rl controllers in dynamic environments,”IEEE Robotics and Automation Letters, vol. 8, no. 11, pp. 7833–7840, 2023

    2023

  24. [32]

    Conformal policy learning for sensorimotor control under distribution shifts,

    H. Huang, S. Sharma, A. Loquercio, A. Angelopoulos, K. Goldberg, and J. Malik, “Conformal policy learning for sensorimotor control under distribution shifts,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 16 285–16 291

    2024

  25. [33]

    Conformal decision theory: Safe autonomous decisions from imperfect predictions,

    J. Lekeufack, A. N. Angelopoulos, A. Bajcsy, M. I. Jordan, and J. Malik, “Conformal decision theory: Safe autonomous decisions from imperfect predictions,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 11 668–11 675

    2024

  26. [34]

    Conformal prediction under covariate shift,

    R. J. Tibshirani, R. Foygel Barber, E. Candes, and A. Ramdas, “Conformal prediction under covariate shift,”Advances in neural information processing systems, vol. 32, 2019

    2019

  27. [35]

    Adaptive conformal prediction for motion planning among dynamic agents,

    A. Dixit, L. Lindemann, S. X. Wei, M. Cleaveland, G. J. Pappas, and J. W. Burdick, “Adaptive conformal prediction for motion planning among dynamic agents,” inLearning for Dynamics and Control Conference. PMLR, 2023, pp. 300–314

    2023

  28. [36]

    Towards generalizable safety in crowd navigation via conformal uncertainty handling,

    J. Yao, X. Zhang, Y . Xia, Z. Wang, A. K. Roy-Chowdhury, and J. Li, “Towards generalizable safety in crowd navigation via conformal uncertainty handling,”arXiv preprint arXiv:2508.05634, 2025

    2025 arXiv

  29. [37]

    Safe reinforcement learning with nonlinear dynamics via model predictive shielding,

    O. Bastani, “Safe reinforcement learning with nonlinear dynamics via model predictive shielding,” in2021 American control conference (ACC). IEEE, 2021, pp. 3488–3494

    2021

  30. [38]

    Probabilistic model predictive safety certification for learning-based control,

    K. P. Wabersich, L. Hewing, A. Carron, and M. N. Zeilinger, “Probabilistic model predictive safety certification for learning-based control,”IEEE Transactions on Automatic Control, vol. 67, no. 1, pp. 176–188, 2021

    2021

  31. [39]

    An efficient spatial-temporal trajectory planner for autonomous vehicles in unstructured environments,

    Z. Han, Y . Wu, T. Li, L. Zhang, L. Pei, L. Xu, C. Li, C. Ma, C. Xu, S. Shenet al., “An efficient spatial-temporal trajectory planner for autonomous vehicles in unstructured environments,”IEEE Transactions on Intelligent Transportation Systems, vol. 25, no. 2, pp. 1797–1814, 2023

    2023

  32. [40]

    Aggressive driving with model predictive path integral control,

    G. Williams, P. Drews, B. Goldfain, J. M. Rehg, and E. A. Theodorou, “Aggressive driving with model predictive path integral control,” in2016 IEEE international conference on robotics and automation (ICRA). IEEE, 2016, pp. 1433–1440

    2016

  33. [41]

    Yolov11: An overview of the key architectural enhancements,

    R. Khanam and M. Hussain, “Yolov11: An overview of the key architectural enhancements,”arXiv preprint arXiv:2410.17725, 2024

    2024 arXiv

  34. [42]

    Bytetrack: Multi-object tracking by associating every detection box,

    Y . Zhang, P. Sun, Y . Jiang, D. Yu, F. Weng, Z. Yuan, P. Luo, W. Liu, and X. Wang, “Bytetrack: Multi-object tracking by associating every detection box,” inEuropean conference on computer vision. Springer, 2022, pp. 1–21

    2022

Pith tools

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