Pith. sign in

REVIEW 3 major objections 5 minor 45 references

The Constitutional Controller: Doubt-Calibrated Steering of Compliant Agents

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A drone controller that folds its own learned position-error distribution into its rule-compliance cost field avoided all collisions in 45 real flights, while a rule-only baseline crashed 11 times.

desk verdict New and promising idea with a real drone experiment, but Equation (10) as written does not type-check and the central mechanism needs a corrected convolution. read the letter →

arxiv 2507.15478 v1 pith:MGCFZ4CJ submitted 2025-07-21 cs.RO cs.AIcs.LG

Simon Kohaut , Felix Divo , Navid Hamid , Benedict Flade , Julian Eggert , Devendra Singh Dhami , Kristian Kersting This is my paper · ORCID
classification cs.ROcs.AIcs.LG
keywords constitutionalcontrollerneuro-symboliccontrolprobabilisticlogicprogrammingself-doubtconditionalnormalizingflowsUAVcomplianceuncertainty-awarepathplanningstatisticalrelationalmaps
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 aims to show that a robot can be steered toward rules-compliance more safely if its planner explicitly accounts for how much the robot expects to miss its commanded position. It introduces the Constitutional Controller (CoCo), which combines a probabilistic-logic constitution of traffic rules with a learned conditional density of the agent's position error. The central move is to replace the raw compliance probability $P(C_t\mid x,z)$ with its expectation under that error density, then plan paths trading off this doubt-calibrated compliance against travel cost. In 45 real drone flights through an obstacle course, CoCo suffered zero crashes while a rule-only baseline crashed on 11 flights, all at high speed. If true, this means uncertainty about one's own control accuracy can be converted directly into safer, still-compliant path choices without changing the rules.

What carries the argument

The load-bearing object is the doubt-calibrated compliance landscape, the expectation of the logic-derived compliance probability under the drone's learned error density. A conditional normalizing flow built from a Masked Autoregressive Transform is fitted to position errors from practice flights, conditioned on controller tuning, speed, and heading; in the cost landscape this density is marginalized out by Monte Carlo, smearing the sharp rule boundaries into safety margins. The final path $\tau^*$ minimizes the summed negative log doubt-calibrated compliance plus a weighted travel-cost term, and an A* planner on a discretized state grid (position and three speed levels) finds that path.

What would settle it

Collect held-out position errors from actual CoCo missions on the obstacle course and compare them with the doubt density fitted on the figure-eight flights; if the density assigns low mass to the large errors that occur near yellow obstacles at 1.0 m/s, then Eq. (10) underestimates the true violation probability and the safety margin would not generalize. Directly retraining the doubt model on the mission course and rerunning the 45-flight comparison is the cleanest test, since the baseline-versus-CoCo gap should shrink if the transferred density was the cause.

Watch

Extended reading notes

Core claim

The paper's central claim is that doubt-calibrated compliance, $P_\theta(C_t\mid x,z)=\int P(C_t\mid x,z)\delta_W(x\mid\theta)\,dx$ in Eq. (10), is what makes a neuro-symbolic controller safe. Here $P(C_t\mid x,z)$ is the probability that the agent's current state and sensor reading satisfy the encoded constitution, and $\delta_W(x\mid\theta)$ is a conditional normalizing flow describing how far the agent expects to be from its commanded state, given speed, heading, and controller tuning. Planning minimizes $\sum_{x\in\tau}[-\alpha \log P_\theta(C_t\mid x,z)+\beta^T J(x)]$ over candidate paths. The empirical result is that this integrated cost steers a small quadcopter away from a short, risky center path at high speeds: the rule-only baseline kept choosing that path and crashed 11 of 45 flights, while CoCo completed all 45 without a crash.

Load-bearing premise

The controller assumes that the position-error statistics learned on practice flights carry over unchanged to the actual obstacle course; if the drone's errors in the mission differ from what it learned, the claimed safety margin is not assured.

Editorial extensions

If this is right

  • A rule-compliant planner gains a built-in safety margin: paths passing close to an obstacle at a speed where the drone's errors are large become less attractive than slower or more distant alternatives.
  • Because setting $\alpha=0$ in Eq. (11) recovers the baseline behavior, CoCo is a strict generalization of the rule-only probabilistic mission planner rather than a separate pipeline.
  • The online compliance probability $P(C_t)$ gives a continuous risk signal during flight, which can trigger emergency routines or post-mission audits in addition to offline path choices.
  • When velocity is left free, CoCo uses its doubt model to slow down over constrained zones and speed up elsewhere, so safety and efficiency are balanced by the same mechanism.

Reading between the lines

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

  • The same convolution trick does not depend on the logic-based constitution: any differentiable cost field could be smoothed by a learned error density, so the idea likely transfers to neural planners and occupancy-grid costs, though the paper does not test that.
  • The transferability of $\delta_W$ is the fragile link: the paper fits doubt on figure-eight flights and applies it to a different course; retraining on the mission course and comparing crash counts would separate the value of doubt-calibration from the value of the specific training distribution.
  • Equation (10) is written as an expectation over $x$ while the left side still depends on $x$; making this well-defined requires specifying the marginalization properly, and a formal treatment of that integral would strengthen the claimed calibration semantics.
  • The observed crash pattern, with all baseline failures at high velocity, suggests the mechanism matters most near the speed where the drone's error distribution widens; adding doubt features such as battery state or wind, as the paper notes as future work, is a natural extension.
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 introduces the Constitutional Controller (CoCo), a neuro-symbolic control architecture for unmanned aerial vehicles. CoCo combines a probabilistic first-order logic program (the Constitution), a Statistical Relational Map (StaR Map) of the environment, sensor/perception clauses, and a learned conditional normalizing flow called the 'doubt density' δW(x|θ). The doubt density is intended to represent the agent's tracking uncertainty under different controller tunings, velocities, and headings. The paper defines a doubt-calibrated compliance landscape Pθ(Ct|x,z) by integrating the compliance landscape with the doubt density, and then plans paths via A* on the cost −α log Pθ(Ct|x,z) + β^T·J(x). The experimental section reports a real-world Crazyflie study: in 45 flights, CoCo had zero crashes while the ProMis baseline crashed 11 times, all at high velocity, and CoCo additionally adapts its velocity when allowed. The paper argues that self-doubt calibration makes compliant agents safer than a baseline that uses the same constitution without the doubt mechanism.

Significance. The paper addresses a relevant problem—rule-compliant and safe autonomous UAV navigation under uncertainty—and the proposed combination of probabilistic logic programs with a learned, context-conditioned error model is genuinely interesting. If the central equations are corrected, the work would provide a concrete neuro-symbolic control mechanism with a real-world demonstration. The experimental effort is a clear strength: 45 physical flights with a Crazyflie, a plausible comparison to a state-of-the-art baseline, and a crash-free result for CoCo. The promise of open-source code is also a positive. However, the formal definition of the doubt-calibrated landscape is currently defective: the central equation, as written, removes the spatial dependence that the downstream planner relies on. Because the entire safety claim rests on this mechanism, the paper needs a substantive revision of the formulation and additional calibration evidence before it can be accepted.

major comments (3)
  1. [Section IV-E, Eq. (10)] Equation (10) does not define the object it claims to define. The left-hand side Pθ(Ct|x,z) retains x, but the right-hand side integrates over x with respect to δW(x|θ); for fixed z and θ, the integral is constant in x. Taken literally, the term −α log Pθ(Ct|x,z) in Eq. (11) cannot distinguish spatial positions, so the planner could not choose different paths based on compliance. This contradicts Figures 7–8, which show doubt-calibrated landscapes that clearly vary with x, and the path-switching behavior reported in Section V-D. The intended construction appears to be a convolution with an error variable, e.g., Pθ(Ct|x,z) = ∫ P(Ct|x+ε,z) δW(ε|θ) dε, but that formula never appears. Moreover, Section IV-D defines δW as a distribution over the robot's state space x, not over deviations from a commanded state. Please state the correct convolution, specify the domain of δW, and give the Monte Carlo estimator that was actually used in the experiments.
  2. [Section IV-D, Eq. (8) and Section V-C, Figure 5] There is a mismatch between the learning objective stated in Eq. (8) and the data described in Section V-C. Eq. (8) maximizes the log-likelihood of 'observed states' x^(i), but Section V-C says the CNF is fit to the L2 norm of the position error between desired and achieved positions, and Figure 5b plots scalar L2 error over time. A scalar L2 error is nonnegative and directionless; it cannot by itself determine the two-dimensional error density shown in Figure 6, and it cannot serve as the convolution kernel needed in Eq. (10). The paper should clarify whether δW(x|θ) models absolute state uncertainty or tracking-error uncertainty, what the actual training target is (2D error vectors or scalar L2 errors), and how the two-dimensional densities in Figure 6 were obtained. This is load-bearing because the doubt-calibrated landscape is only meaningful if the learned object has the right support and dimensionality.
  3. [Section V-C to V-D and Section V-E] The doubt density is trained on 8-shaped flights at speeds shown in Figure 5 (0.6–1.0 m/s) and then applied to a different obstacle course at planned speeds that include 0.2 and 0.5 m/s, which are outside the training range. The paper provides no held-out calibration, coverage analysis, or evidence that the learned error model remains valid near the obstacles and under the actual path-following controller used in the mission. The central safety claim—that doubt-calibrated planning avoids crashes—depends on this transferability. Section V-E only notes that tuning may be context-dependent; it does not address calibration. Please add quantitative calibration checks (e.g., held-out error likelihoods, quantile coverage of the error density, or reliability diagrams) or explicitly scope the claim to conditions covered by the training distribution.
minor comments (5)
  1. [Section V-B] There is a typo in the state description: 'veocity' should be 'velocity'.
  2. [Section V-E] The statement that 'choosing α=0 perfectly recovers the baseline behavior' is inaccurate: with α=0, Eq. (11) ignores the compliance landscape P(Ct|x,z), whereas the ProMis baseline is described as also minimizing a compliance term with weight α=2. Please rephrase the subsumption claim.
  3. [Section IV-F, Eq. (11)] The notation J is overloaded: J(x) ∈ R^J with J ∈ N is confusing. Use a different symbol, e.g., K, for the dimensionality of the cost vector.
  4. [Section IV-G, Eq. (13)] The integral in Eq. (13) uses a filtering density p(xt|z1:t-1) but no online Bayesian update is specified in the paper; either define this quantity explicitly or move the equation to the discussion as a conceptual quantity.
  5. [Figures 7 and 8] The captions say the doubt density is 'marginalized' for velocity or heading, but the notation in Eq. (10) does not show conditioning on θ; please align the notation so that the reader can see which doubt features are being marginalized in each figure.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the doubt model is explicitly learned from data, and the safety advantage is an empirical comparison against a shared baseline, not a consequence of the definitions.

full rationale

The paper's derivation chain is: fit a conditional doubt density δW(x|θ) by maximizing the log-likelihood of measured position errors (Eq. 8); use it to form a doubt-calibrated compliance term Pθ (Eqs. 9-10); and minimize that term together with travel cost in Eq. (11). The doubt density is openly fitted on training flights, not on the later crash outcomes, so this is disclosed supervised learning rather than a prediction that reduces to its inputs. The safety claim is tested empirically: Section V-D compares CoCo against ProMis using the same Constitution, StaR Map, and A* graph, with only the doubt convolution differing, and reports 0 crashes in 45 flights versus 11 in the baseline. That result could have been different if the learned density were miscalibrated, so it is not forced by construction. The self-citations to ProMis [6], StaR Maps [24], and related mission-design work are background components and the baseline; no uniqueness theorem is invoked and no load-bearing conclusion is imported solely from the authors' prior papers. A formal presentation issue exists in Eq. (10), where x appears both as the left-hand argument and as the integration variable, so a literal reading makes Pθ position-independent and unable to affect Eq. (11); this is a notation/definition problem, not a circular reduction, and the reported path switching indicates the implementation uses an unwritten convolution over state deviations.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central method rests on a small number of hand-set probabilities and cost weights, plus assumptions that the learned error model transfers and that the map segmentation is trustworthy. No new physical entities are introduced; the doubt density is a fitted statistical model, not a postulated entity.

free parameters (4)
  • Constitution rule probabilities (0.2 :: over(x,red), 0.1 :: over(x,green), distance(x,yellow) ~ normal(1.0,0.2)… = hand-set values in Listing 1
    These probabilities and distribution parameters define the compliance landscape P(Ct|x,z) and are chosen by the authors, not learned from data.
  • Cost weights α and β = α=2, β=[1]
    Set in Section V-D for the A* objective (Eq 11); the comparison outcome depends on these choices.
  • StaR Map positional uncertainty = predetermined, value not given
    Section V-A says obstacle geometries are modeled by a StaR Map with predetermined positional uncertainty; this affects the width of compliance probabilities.
  • CNF architecture and training hyperparameters = 5 flow layers, 100 hidden units, Adam optimizer
    Architecture choices in Section V-C; they shape the learned doubt density but are not fitted by the proposed method.
assumptions (5)
  • domain assumption Ground atoms in the constitution are independent enough for the sum-product exact inference in Eq (7).
    Section IV-C multiplies atom probabilities without stating an independence assumption or program structure that justifies the factorization.
  • domain assumption The learned doubt density δW(x|θ) is stationary and transfers from training trajectories to the mission environment.
    Section V-C fits the density on 8-shaped flights; Section V-D applies it to a different obstacle course without calibration on held-out missions.
  • domain assumption The doubt density is a distribution over the full state (or position) used in Eq (10), not merely a scalar L2 error.
    Experiments say the fitted quantity is the L2 norm of position error, while Eq (10) treats it as a density over x. The integration variable is ambiguous.
  • domain assumption A* search on a 300x300x3 grid with velocities 0.2, 0.5, 1.0 m/s captures the relevant path costs.
    Section V-D uses this discretization; planning is approximate, and CoCo's dynamic speed adaptation is restricted to three levels.
  • domain assumption Map and obstacle segmentation from SAM2 is accurate enough for the StaR Map.
    Section V-A relies on SAM2 segmentations; no error analysis of the segmentation is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Constitutional Controller: Doubt-Calibrated Steering of Compliant Agents." pith.science (2026). https://pith.science/paper/MGCFZ4CJ

@misc{pith2026250715478,
  author       = {Pith},
  title        = {Pith review of: The Constitutional Controller: Doubt-Calibrated Steering of Compliant Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MGCFZ4CJ}},
  note         = {Machine review of arXiv:2507.15478}
}
read the original abstract

Ensuring reliable and rule-compliant behavior of autonomous agents in uncertain environments remains a fundamental challenge in modern robotics. Our work shows how neuro-symbolic systems, which integrate probabilistic, symbolic white-box reasoning models with deep learning methods, offer a powerful solution to this challenge. This enables the simultaneous consideration of explicit rules and neural models trained on noisy data, combining the strength of structured reasoning with flexible representations. To this end, we introduce the Constitutional Controller (CoCo), a novel framework designed to enhance the safety and reliability of agents by reasoning over deep probabilistic logic programs representing constraints such as those found in shared traffic spaces. Furthermore, we propose the concept of self-doubt, implemented as a probability density conditioned on doubt features such as travel velocity, employed sensors, or health factors. In a real-world aerial mobility study, we demonstrate CoCo's advantages for intelligent autonomous systems to learn appropriate doubts and navigate complex and uncertain environments safely and compliantly.

Figures

Figures reproduced from arXiv: 2507.15478 by the authors.

Figure 1
Figure 1. Constitutional Control for navigating complex, regulated spaces: Based on neuro-symbolic reasoning over the compliant airspaces, CoCo integrates its learned doubts about the agent’s capabilities to steer its actions towards safety and compliance. substantially harmonize and advance UAV legislation. While AAM was not considered in the initial Master Plan in 2009, with drones merely mentioned in passing, the subsequen… view at source ↗
Figure 2
Figure 2. StaR Maps parameters in an Advanced Aerial Mobility setting: Here, the expected values of two basic probabilistic spatial relations are shown in an urban environment, namely over and distance (compare [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The Constitutional Controller’s architecture: CoCo rea￾sons on a neuro-symbolic model of the environment’s rules while expressing its learned doubt model on the Constitution’s likelihood, integrating background knowledge, perception, and a probabilistic environment representation into the control process. f(ψ) with parameters ψ if its body is true. If the right-hand side is empty, the head is regarded as a fact and … view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Applying the Constitutional Controller to a real-world aviation testing environment: In our testbed, an autonomous agent is set up to travel from one of the blue vertiports to the other. For its journey, the agent can take one of three paths: A short but risky one thro…
Figure 6
Figure 6. Figure 6: Learning to doubt: We show the doubt density fitted on test flight data as presented in [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 5
Figure 5. Figure 5: Gathering doubt training-data: We run 8-shaped test flights of the UAV (a) to train a doubt density of the error (b) as Conditional Normalizing Flow conditioned on controller parameters, speed, and heading angle. The target trajectory is shown in black, with the actual…
Figure 7
Figure 7. Figure 7: Doubt expression on the navigation space: After fitting the Conditional Flow on the UAV’s test flights (a), we show the landscapes Pθ(Ct|x, z) with the doubt density marginalized for the agent’s velocity (b–d). v = 0.2 ms − 1 v = 0.5 ms − 1 v = 1.0 ms − 1 (a) Pθ(Ct|x, …
Figure 8
Figure 8. Figure 8: Affecting the compliance landscape through doubt features: We exemplify the impact of varying doubt features in two ways. First, with a fixed controller tuning and heading angle, one can see how the landscape is smeared according to the doubt density and changes due to…
Figure 9
Figure 9. Figure 9: Constitutional Control for safe and compliant agents: Because CoCo learned the UAV’s capabilities, it dynamically switches paths for different target speeds. In contrast, the baseline ProMis always greedily routes through the center path, crashing (red ✕) into the yell…
Figure 10
Figure 10. Figure 10: Compliance assurance with Constitutional Control: Here, we show how CoCo assures a high probability of compliance compared to the baseline, both in the case of a constrained (a) and optimized (b) velocity. Compared to the baseline, which does not consider the change i…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 41 canonical work pages

  1. [1]

    Incremental reasoning in probabilistic signal temporal logic,

    M. Tiger and F. Heintz, “Incremental reasoning in probabilistic signal temporal logic,” International Journal of Approximate Reasoning , vol. 119, pp. 325–352, 2020. [Online]. Available: https://www.sciencedirect. com/science/article/pii/S0888613X19302671

    work page 2020

  2. [2]

    European atm master plan: The roadmap for delivering high performing aviation for europe: Executive view: Edition 2015,

    SESAR Joint Undertaking, “European atm master plan: The roadmap for delivering high performing aviation for europe: Executive view: Edition 2015,” 2015, [Online]

    work page 2015

  3. [3]

    European drones outlook study,

    ——, “European drones outlook study,” 2017, [Online]

    work page 2017

  4. [4]

    Advanced air mobility: Opportunities and challenges deploying evtols for air ambulance service,

    R. Goyal and A. Cohen, “Advanced air mobility: Opportunities and challenges deploying evtols for air ambulance service,”Applied Sciences, vol. 12, no. 3, p. 1183, 2022

    work page 2022

  5. [5]

    Bibliometric analysis on advanced air mobility and drones,

    C. Purtell, S.-J. Hong, and B. Hiatt, “Bibliometric analysis on advanced air mobility and drones,” Journal of Air Transport Management, vol. 116, p. 102569, 2024. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0969699724000346

    work page 2024

  6. [6]

    Mission design for unmanned aerial vehicles using hybrid probabilistic logic programs,

    S. Kohaut, B. Flade, D. S. Dhami, J. Eggert, and K. Kersting, “Mission design for unmanned aerial vehicles using hybrid probabilistic logic programs,” in 2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC) . IEEE, 2023, pp. 1506–1513

    work page 2023

  7. [7]

    The Constitutional Filter: Bayesian Estimation of Compliant Agents

    S. Kohaut, F. Divo, B. Flade, D. S. Dhami, J. Eggert, and K. Kersting, “The constitutional filter,” arXiv preprint arXiv:2412.18347 , 2024

    work page Pith review arXiv 2024

  8. [8]

    Probabilistic mission design in neuro-symbolic systems,

    S. Kohaut, B. Flade, D. Ochs, D. S. Dhami, J. Eggert, and K. Kersting, “Probabilistic mission design in neuro-symbolic systems,”

    work page

Show all 45 references
  1. [9]

    An introduction to prolog III,

    A. Colmerauer, “An introduction to prolog III,” Communications of the ACM, vol. 33, no. 7, pp. 69–90, 1990

    1990

  2. [10]

    Bayesian logic programming: Theory and tool,

    K. Kersting and L. De Raedt, “Bayesian logic programming: Theory and tool,” Introduction to Statistical Relational Learning , p. 291, 2007

    2007

  3. [11]

    Problog: A probabilistic prolog and its application in link discovery

    L. De Raedt, A. Kimmig, and H. Toivonen, “Problog: A probabilistic prolog and its application in link discovery.” in IJCAI, vol. 7. Hyder- abad, 2007, pp. 2462–2467

    2007

  4. [12]

    Inference and learning in probabilistic logic programs using weighted boolean formulas,

    D. Fierens, G. Van den Broeck, J. Renkens, D. Shterionov, B. Gutmann, I. Thon, G. Janssens, and L. De Raedt, “Inference and learning in probabilistic logic programs using weighted boolean formulas,” Theory and Practice of Logic Programming , vol. 15, no. 3, pp. 358–401, 2015

    2015

  5. [13]

    DeepProbLog: Neural probabilistic logic programming,

    R. Manhaeve, S. Dumancic, A. Kimmig, T. Demeester, and L. De Raedt, “DeepProbLog: Neural probabilistic logic programming,” advances in neural information processing systems , vol. 31, 2018

    2018

  6. [14]

    Neurasp: Embracing neural networks into answer set programming,

    Z. Yang, A. Ishay, and J. Lee, “Neurasp: Embracing neural networks into answer set programming,” in 29th International Joint Conference on Artificial Intelligence (IJCAI 2020) , 2020

    2020

  7. [15]

    Scalable neural- probabilistic answer set programming,

    A. Skryagin, D. Ochs, D. S. Dhami, and K. Kersting, “Scalable neural- probabilistic answer set programming,” Journal of Artificial Intelligence Research, vol. 78, pp. 579–617, 2023

    2023

  8. [16]

    Towards probabilistic clearance, explanation and optimization,

    S. Kohaut, B. Flade, D. S. Dhami, J. Eggert, and K. Kersting, “Towards probabilistic clearance, explanation and optimization,” in 2024 Inter- national Conference on Unmanned Aircraft Systems (ICUAS) . IEEE, 2024, pp. 911–916

    2024

  9. [17]

    Probabilistic logic programming for hybrid relational domains,

    D. Nitti, T. De Laet, and L. De Raedt, “Probabilistic logic programming for hybrid relational domains,” Machine Learning , vol. 103, pp. 407– 449, 2016

    2016

  10. [18]

    Probabilistic robotics,

    S. Thrun, “Probabilistic robotics,” Communications of the ACM, vol. 45, no. 3, pp. 52–57, 2002

    2002

  11. [19]

    Probability map building of uncertain dynamic environments with indistinguishable obstacles,

    M. Jun and R. D’Andrea, “Probability map building of uncertain dynamic environments with indistinguishable obstacles,” in Proceedings of the 2003 American Control Conference, 2003. , vol. 4. IEEE, 2003, pp. 3417–3422

    2003

  12. [20]

    Improving grid-based slam with rao-blackwellized particle filters by adaptive proposals and selective resampling,

    G. Grisetti, C. Stachniss, and W. Burgard, “Improving grid-based slam with rao-blackwellized particle filters by adaptive proposals and selective resampling,” in Proceedings of the 2005 IEEE international conference on robotics and automation . IEEE, 2005, pp. 2432–2437

    2005

  13. [21]

    Dynamic environment modeling with gridmap: a multiple-object tracking application,

    C. Chen, C. Tay, C. Laugier, and K. Mekhnacha, “Dynamic environment modeling with gridmap: a multiple-object tracking application,” in 2006 9th International Conference on Control, Automation, Robotics and Vision. IEEE, 2006, pp. 1–6

    2006

  14. [22]

    Lidar- based perception for offroad navigation,

    M. Himmelsbach, F. von Hundelshausen, and H.-J. W ¨unsche, “Lidar- based perception for offroad navigation,” in 3. Tagung Aktive Sicherheit durch Fahrerassistenz, 2008

    2008

  15. [23]

    Scalar field mapping with adaptive high-intensity region avoidance,

    M. Qureshi, T. E. Ogri, Z. I. Bell, and R. Kamalapurkar, “Scalar field mapping with adaptive high-intensity region avoidance,” in 2024 IEEE Conference on Control Technology and Applications (CCTA) . IEEE, 2024, pp. 388–393

    2024

  16. [24]

    Star maps: Unveiling uncertainty in geospatial relations,

    B. Flade, S. Kohaut, J. Eggert, D. S. Dhami, and K. Kersting, “Star maps: Unveiling uncertainty in geospatial relations,” in 2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC) . IEEE, 2024

    2024

  17. [25]

    Introduction to kalman filter and its applica- tions,

    Y . Kim, H. Bang et al. , “Introduction to kalman filter and its applica- tions,” Introduction and Implementations of the Kalman Filter , vol. 1, pp. 1–16, 2018

    2018

  18. [26]

    A new approach to linear filtering and prediction problems,

    R. E. Kalman, “A new approach to linear filtering and prediction problems,” Journal of Basic Engineering , vol. 82, no. 1, pp. 35–45, 03

  19. [27]

    A new approach for filtering nonlinear systems,

    S. J. Julier, J. K. Uhlmann, and H. F. Durrant-Whyte, “A new approach for filtering nonlinear systems,” in Proceedings of 1995 American Control Conference-ACC’95, vol. 3. IEEE, 1995, pp. 1628–1632

    1995

  20. [28]

    New extension of the kalman filter to nonlinear systems,

    S. J. Julier and J. K. Uhlmann, “New extension of the kalman filter to nonlinear systems,” in Signal processing, sensor fusion, and target recognition VI, vol. 3068. Spie, 1997, pp. 182–193

    1997

  21. [29]

    A literature review of uav 3d path planning,

    L. Yang, J. Qi, J. Xiao, and X. Yong, “A literature review of uav 3d path planning,” in Proceeding of the 11th world congress on intelligent control and automation . IEEE, 2014, pp. 2376–2381

    2014

  22. [30]

    A formal basis for the heuristic determination of minimum cost paths,

    P. E. Hart, N. J. Nilsson, and B. Raphael, “A formal basis for the heuristic determination of minimum cost paths,” IEEE transactions on Systems Science and Cybernetics , vol. 4, no. 2, pp. 100–107, 1968

    1968

  23. [31]

    Implementing 3d network analysis in 3d-gis,

    I. A. Musliman, A. A. Rahman, V . Coors et al. , “Implementing 3d network analysis in 3d-gis,” International archives of ISPRS , vol. 37, no. part B, 2008

    2008

  24. [32]

    Real-time continuous curvature path plan- ning of uavs in cluttered environments,

    K. Yang and S. Sukkarieh, “Real-time continuous curvature path plan- ning of uavs in cluttered environments,” in 2008 5th international symposium on mechatronics and its applications . IEEE, 2008, pp. 1–6

    2008

  25. [33]

    Hybrid evolutionary approach to multi-objective path planning for UA Vs,

    N. Hohmann, M. Bujny, J. Adamy, and M. Olhofer, “Hybrid evolutionary approach to multi-objective path planning for UA Vs,” in Proc SSCI . IEEE, 2021, pp. 1–8

    2021

  26. [34]

    Multi-objective 3D path planning for UA Vs in large-scale urban scenarios,

    ——, “Multi-objective 3D path planning for UA Vs in large-scale urban scenarios,” in Proc CEC. IEEE, 2022, pp. 1–8

    2022

  27. [35]

    Swi-prolog,

    J. Wielemaker, T. Schrijvers, M. Triska, and T. Lager, “Swi-prolog,” Theory and Practice of Logic Programming , vol. 12, no. 1-2, pp. 67– 96, 2012

    2012

  28. [36]

    Multi-shot ASP solving with clingo,

    M. Gebser, R. Kaminski, B. Kaufmann, and T. Schaub, “Multi-shot ASP solving with clingo,” CoRR, vol. abs/1705.09811, 2017

    2017 arXiv

  29. [37]

    Masked autoregressive flow for density estimation,

    G. Papamakarios, T. Pavlakou, and I. Murray, “Masked autoregressive flow for density estimation,” Advances in neural information processing systems, vol. 30, 2017

    2017

  30. [38]

    Crazyflie 2.1: Quadcopter,

    Bitcraze, “Crazyflie 2.1: Quadcopter,” https://www.bitcraze.io/products/ old-products/crazyflie-2-1/, accessed: 2025-02-26

    2025

  31. [39]

    Crazyswarm: A large nano-quadcopter swarm,

    J. A. Preiss, W. Honig, G. S. Sukhatme, and N. Ayanian, “Crazyswarm: A large nano-quadcopter swarm,” in 2017 IEEE International Confer- ence on Robotics and Automation (ICRA). IEEE, 2017, pp. 3299–3304

    2017

  32. [40]

    Controllers in the crazyflie,

    Bitcraze, “Controllers in the crazyflie,” https://www.bitcraze.io/ documentation/repository/crazyflie-firmware/master/functional-areas/ sensor-to-control/controllers/, accessed: 2025-01-27

    2025

  33. [41]

    Valve index base station,

    SteamVR, “Valve index base station,” https://store.steampowered.com/ app/1059570/Valve Index Basisstation/?l=english, accessed: 2025-02- 26

    2025

  34. [42]

    SAM 2: Segment anything in images and videos,

    N. Ravi, V . Gabeur, Y .-T. Hu, R. Hu, C. Ryali, T. Ma, H. Khedr, R. R ¨adle, C. Rolland, L. Gustafson, E. Mintun, J. Pan, K. V . Alwala, N. Carion, C.-Y . Wu, R. Girshick, P. Dollar, and C. Feichtenhofer, “SAM 2: Segment anything in images and videos,” in The Thirteenth Inter...

    2025

  35. [1960]

    Available: https://doi.org/10.1115/1.3662552

    [Online]. Available: https://doi.org/10.1115/1.3662552

    doi

  36. [2022]

    Kristian Kersting is a Full Professor at the Com- puter Science Department of the TU Darmstadt University, Germany

    Devendra’s research interests currently focus on successfully incorporating causality and reason- ing into deep learning systems. Kristian Kersting is a Full Professor at the Com- puter Science Department of the TU Darmstadt University, Germany. He is the head of the Artificia...

    2006

  37. [2024]

    Available: https://arxiv.org/abs/2501.01439

    [Online]. Available: https://arxiv.org/abs/2501.01439

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.