Pith. sign in

REVIEW 4 major objections 5 minor 37 references

Navigating the Proximity-Safety Balance: Constraint Decomposition for Human Following in Pedestrian Crowds

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Decomposing robot following into a sparse reward and three cost constraints gives explicit control over the proximity-safety trade-off.

desk verdict A solid systems paper with a real empirical payload, but the following-cost dead zone in Eq. (5) breaks the paper's central claim that thresholds provide direct behavioral control. read the letter →

arxiv 2608.10056 v1 pith:RFG6LOK7 submitted 2026-08-10 cs.RO cs.AIcs.LGcs.SYeess.SY

Shiting Gong , Jianpeng Yao , Jinfeng Wang , Marco Pavone , Jiachen Li This is my paper · ORCID
classification cs.ROcs.AIcs.LGcs.SYeess.SY
keywords human-followingrobotconstrainedreinforcementlearningcostconstraintsproximity-safetytrade-offadaptiveconformalinferencepedestriancrowdsPPO-Lagrangianuncertainty-awarenavigation
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 argues that the conflict between staying close to a target human and avoiding collisions in crowds is best managed by separating the task into a sparse reward for following and three independent cost constraints for following distance, human safety, and obstacle safety. Instead of tuning reward weights, the designer sets cost thresholds—a desired following distance, a tolerable intrusion level, and a safe obstacle clearance—that map directly onto behavior. The paper shows in simulation that shifting these thresholds moves the proximity-safety balance in the intended direction, while doubling reward weights can leave behavior unchanged or make it worse. It also feeds online prediction-uncertainty estimates into the human-safety cost, so the robot keeps a wider buffer when pedestrian motion is hard to predict. If correct, the framework gives robot designers a verifiable way to set how aggressively a follower robot should behave in dense, unfamiliar crowds.

What carries the argument

The load-bearing object is the constraint decomposition itself: the CMDP in Eq. (3) with reward $R_t$ and costs $C^F_t$, $C^H_t$, $C^O_t$ subject to thresholds $\delta_F$, $\delta_H$, $\delta_O$, optimized via PPO-Lagrangian with four critics and a combined advantage. The supporting mechanism is the adaptive conformal inference bound $\hat\delta_{h,k}(t)$ from Eq. (2), which is inserted both into the observation tokens and into the human-safety cost, where it expands the radius of safety regions around predicted pedestrian positions by exactly the current error bound. The attention-based Transformer policy fuses robot, target, obstacle, and human tokens so spatial constraints and social interactions are reasoned over jointly; at inference only the actor remains, so the decomposition costs nothing extra at deployment.

What would settle it

Run the trained policy in a crowd where pedestrians make abrupt 90-degree direction changes and measure, over all prediction horizons, how often the true prediction error exceeds the adaptive conformal bound; if the empirical coverage falls well below the nominal $1-\alpha$ level, the safety cost has stopped encoding collision risk and the claimed safety behavior is not guaranteed.

Watch

Extended reading notes

Core claim

The central claim is that in a constrained Markov decision process, a sparse task reward plus three independent cost constraints with behaviorally meaningful thresholds yields a proximity-safety balance that is explicit, tunable, and more predictable than the balance produced by a dense reward with tuned weights. The following cost uses an equality constraint to hold a desired distance $d_{personal}$, while human and obstacle collision costs use inequality constraints with limits $\delta_H$ and $\delta_O$; each cost is evaluated by its own critic and the policy is trained with PPO-Lagrangian. Prediction uncertainty from an adaptive conformal inference module, written $\hat\delta_{h,k}(t)$ for human $h$ at horizon $k$, expands the safety region around each predicted position when prediction errors grow. Direct comparisons show that changing $\delta_F$ and $\delta_H$ shifts average following distance and collision rates in the expected directions, whereas doubling reward weights fails to produce the intended effect and can degrade both following and safety.

Load-bearing premise

The human-safety cost carries collision risk only if the uncertainty estimates around predicted pedestrian positions truly bound the real prediction errors under the crowd behaviors encountered, including sudden turns and rushing pedestrians.

Editorial extensions

If this is right

  • A designer can specify the desired behavior directly by setting $\delta_F$, $\delta_H$, and $\delta_O$: lower $\delta_F$ gives closer following, lower $\delta_H$ gives safer human interaction, and each moves the other in the expected direction.
  • Because thresholds map predictably to metrics, a designer can pre-select the intended behavior in simulation and then deploy the same thresholds, rather than hoping tuned weights generalize.
  • The uncertainty-aware cost extends to out-of-distribution crowds such as bidirectional corridors, rushing pedestrians, social-force groups, and moving groups, where the method keeps collision rates below non-uncertainty baselines.
  • At inference the actor network alone is used, so the explicit trade-off control comes with no additional runtime cost compared with single-critic RL.
  • Reward-weight tuning can fail in a counterintuitive way, as doubling the following weight left following distance unchanged and raised both target-loss and collision rates; cost thresholds avoid this failure mode.

Reading between the lines

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

  • Editorial inference: the same decomposition could be applied to other navigation or mobile-manipulation tasks with competing objectives, such as social navigation with personal-space preferences, replacing weight ratios with thresholds.
  • Editorial inference: because the ACI module only needs observed prediction errors, a more accurate pedestrian predictor should shrink the uncertainty buffers and permit closer following at the same safety level, which is a testable prediction.
  • Editorial inference: the equality constraint on the following cost may make training infeasible when $\delta_F$ is set below what the robot's dynamics allow; the paper reports tuned values, but mapping the feasible range of $\delta_F$ would clarify how far the tuning promise extends.
  • Editorial inference: the real-robot failures were attributed to upstream perception rather than the policy, so a testable extension is to feed detection confidence into the same uncertainty-aware cost structure.
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

4 major / 5 minor

Summary. The paper proposes a constrained reinforcement learning (CRL) framework for human-following robots in crowded pedestrian environments. The task is decomposed into a sparse task reward and three independent cost constraints—following distance, human safety, and obstacle safety—with thresholds δ_F, δ_H, and δ_O intended to have direct behavioral meaning. Prediction uncertainty of human motion is quantified via adaptive conformal inference (ACI) and integrated into both the observation space and the human-safety cost. The method is evaluated in an extended CrowdNav simulator with static obstacles, across in-distribution and out-of-distribution scenarios, and deployed on a real robot. The central claim is that cost-threshold tuning provides explicit and predictable control over the proximity-safety trade-off, in contrast to implicit reward-weight tuning.

Significance. If the threshold-control claim held, the framework would offer a practical and interpretable alternative to dense reward shaping in human-following navigation, and the integration of ACI-based uncertainty into CRL costs is a sensible and potentially useful design. The paper contains substantial experimental work: multiple classical and learning baselines, four OOD scenarios, ablations, and a real-robot deployment. However, the central mechanism for the following-distance constraint is flawed as written (Eq. (5) has a dead zone below d_personal), and the experimental reporting is weakened by the selection of the 'best result among all tuned configurations' and the absence of variance statistics despite five seeds. The uncertainty-integration contribution also overlaps heavily with the authors' prior work [20,22]. With revisions, the framework could become a solid contribution, but the current manuscript does not substantiate its central interpretability claim.

major comments (4)
  1. [III-D, Eq. (5)] The following cost is defined as C_F = k1(d_follow - d_personal) for d_follow > d_personal and 0 otherwise, so distances below d_personal incur no penalty. The equality constraint E[sum_t C_F] = δ_F can therefore be satisfied by a policy that spends some time arbitrarily close to the target and some time far away, with no cost signal preventing personal-space intrusion. The claim in Section IV.F that 'Tightening δ_F produces closer following' is not supported by this cost definition; at most δ_F bounds the average excess distance above d_personal. The manuscript should either modify C_F to penalize both sides of the desired distance (or add a separate personal-space constraint) and rerun the experiments, or substantially weaken the interpretability claims for δ_F.
  2. [IV.C and IV.D] Section IV.C states that for both RL+ACI and the proposed method the authors 'report the best result among all tuned configurations,' and Section IV.D says 1250 test samples across 5 seeds are used. Tables I–III report only point estimates with no standard deviations or error bars. As a result, the claimed improvements (e.g., SR 78.08% vs 71.60% for RL+ACI in Table I, and the monotonic trends in Table II) cannot be separated from tuning luck and seed variance. Please report mean ± std (or confidence intervals) across seeds for all metrics, disclose the full set of tuned configurations, and state how 'best' was chosen; otherwise the experimental comparisons are not statistically interpretable.
  3. [III-D and IV.G] The human-safety cost C_H is built on ACI uncertainty bounds δ̂_{h,k}(t), and the paper claims improved safety in unpredictable OOD scenarios. However, no empirical coverage statistics for the ACI bounds are reported for the OOD conditions (rushing humans, SF model, groups, corridor), where the exchangeability assumptions underlying conformal prediction are violated. Without coverage or calibration data, the interpretation of δ_H as a safety threshold with direct behavioral meaning is not verified. Please report per-scenario empirical coverage, average error-bound ratios, and, if possible, conditional coverage as a function of horizon and crowd density.
  4. [III-C,D and Contributions] The uncertainty-aware cost formulation is listed as a main contribution, but Refs [20] and [22], both from the same research group, already present ACI-based uncertainty bounds integrated into constrained RL costs for crowd navigation. The manuscript should explicitly differentiate the present contribution from these works (e.g., the human-following setting, the interaction between the following constraint and the uncertainty-aware costs, or any architectural changes). As written, the novelty of the uncertainty-integration component is overstated, and the relation to prior self-authored work should be clarified in both the related work and the contributions list.
minor comments (5)
  1. [Eq. (2)] The indicator notation in Eq. (2) is easy to misread; please define the indicator explicitly (e.g., I[·]) and clarify the dependence on h and k.
  2. [Tables I and III] The column grouping under 'CR' should be labeled explicitly (e.g., 'CR (Overall / Human / Obstacle)') to avoid ambiguity.
  3. [IV.H] Section IV.H reports 10 real-robot trajectories with 7 successes but does not state the criteria for success, the number of moving pedestrians, or the specific perception failures; a short table or bullet list would help substantiate the feasibility claim.
  4. [Section IV.D] The paper does not mention whether code and simulation environments will be released; for reproducibility, please add a statement or public repository link on the project page.
  5. [Eq. (12)] In Eq. (12), λ_F is updated without a non-negativity clamp, unlike Eq. (11); please clarify whether negative multipliers are allowed and how the equality constraint is enforced in practice.

Circularity Check

1 steps flagged · score 6.0 of 10

Following-cost constraint is a hinge-loss budget above d_personal, so the claimed 'desired following distance' control is not encoded; uncertainty-cost self-citations are not load-bearing.

  1. self definitional [Section III-D, Eq. (5) and the constraint formulation in Eq. (3); discussed in Section IV.F]
    "the following cost uses an equality constraint, maintaining a desired distance rather than minimizing it to avoid both target loss and personal-space intrusion. ... C_F_t = ( k1(d_follow,t − d_personal ), if d_follow,t > d_personal , 0, otherwise ) (5)"

    By Eq. (5), C_F=0 for all d_follow < d_personal, so the equality constraint E[Σ_t C_F]=δ_F in Eq. (3) is by construction a budget on the time-integrated positive excess distance above d_personal, not on the following distance itself and not on personal-space intrusion. The paper's claimed behavioral meaning for δ_F—'a desired following distance', 'moderate following distance rather than crowding'—is therefore not encoded in the cost; a policy can sit at zero distance at zero cost and satisfy δ_F by occasionally being far away. The 'explicit control' over the proximity side of the trade-off reduces to setting a hinge-loss budget; the crowding-avoidance component is imported by assertion.

full rationale

The paper's main derivation—training a PPO-Lagrangian policy under three cost constraints with thresholds δ_F, δ_H, δ_O—is not circular with respect to its empirical claims: the thresholds are set a priori, the costs are specified, and the behavior is measured on held-out scenarios against external baselines. The ACI uncertainty estimator in Eq. (2) is taken from external conformal-prediction literature (Gibbs & Candès; Lindemann et al.), and the present paper provides its own experiments, so the self-citations [20] and [22] are not load-bearing. The central circularity concern is confined to the following-distance constraint. Eq. (5) defines C_F as the positive part of (d_follow − d_personal); hence Eq. (3)'s equality constraint controls only the time-integrated excess above d_personal. The paper's repeated claim that δ_F encodes a 'desired following distance' or prevents 'personal-space intrusion' is not a consequence of the defined cost; that behavioral meaning is asserted, not derived. Table II's observation that decreasing δ_F lowers AFD is consistent with this hinge-loss budget, but the 'moderate distance / no crowding' component is not forced by the constraint, and the Lagrangian update provides no gradient in the dead zone. Additionally, Section IV.C admits that both Ours and RL+ACI report 'the best result among all tuned configurations,' which is a comparison-selection limitation, but it is not circularity. Overall, one of the three threshold meanings reduces by construction to a budget on excess distance, making the proximity-control claim partially circular; the safety-threshold claims and the empirical comparisons retain independent content.

Assumptions & free parameters 11 free parameters · 5 assumptions · 0 invented entities

The central claims rely on many hand-set thresholds and unreported coefficients, including k1, k2, k3, delta limits, sparse reward constants, and ACI hyperparameters, plus the assumption that ACI bounds and ORCA simulation transfer to real crowds. None of these are derived from first principles, and the paper does not provide code to inspect them.

free parameters (11)
  • k1 (following cost coefficient) = not reported
    Scales C_F in Eq. (5); not specified in the paper, but it determines how delta_F translates into following distance.
  • k2 (human intrusion cost coefficient) = not reported
    Scales C_H in Eq. (6); not specified, affects the safety versus following balance.
  • k3 (obstacle cost coefficient) = not reported
    Scales C_O in Eq. (7); not specified.
  • delta_F (following cost limit) = 3.6 in balanced profile; 4.0 and 3.2 in other profiles
    Cost limit for following; manually tuned in Table II to shift target lost rate versus human collision rate.
  • delta_H (human safety cost limit) = 3.6 in balanced profile; 3.2 and 4.0 in other profiles
    Cost limit for human safety; manually tuned across profiles.
  • delta_O (obstacle safety cost limit) = 1.2 (fixed)
    Fixed across compared configurations; no derivation from robot dynamics is provided.
  • d_valid (valid following distance) = 5.0 m
    Defines target loss threshold and perception range; application-specific choice.
  • d_personal (personal-space threshold) = 1.0 m
    Determines when the following cost activates; hand-set.
  • d_safe_o (obstacle safe distance) = 0.50 m
    Hand-set obstacle safety threshold in Eq. (7).
  • ACI hyperparameters (M estimators, alpha, gamma_m, sampling rule) = not reported
    Required by Eq. (2); values are absent, yet they control the uncertainty bounds used in the human safety cost.
  • Sparse reward constants (R_success, R_collision, R_target_lost) = not reported
    Values in Eq. (4) are never given; they set the scale of the task reward relative to the Lagrangian cost advantages.
assumptions (5)
  • domain assumption ACI online error bounds remain valid under arbitrary pedestrian distribution shifts.
    Eq. (2) updates bounds using observed errors; the human safety cost in Section III-D expands regions by these bounds, so if coverage fails under sudden turns or rushing pedestrians, C_H no longer measures true collision risk.
  • domain assumption ORCA-based crowds in CrowdNav are representative of real dense pedestrian behavior for training.
    Training and in-distribution evaluation use ORCA-controlled agents; real-robot transfer assumes the simulated distribution is close enough to deployment conditions.
  • domain assumption PPO-Lagrangian dual descent converges to a feasible policy for the three-constraint CMDP.
    The method relies on the standard Lagrangian learning loop in Eqs. (9) through (12); non-convex deep RL gives no convergence guarantee, and no constraint-satisfaction curves are shown.
  • domain assumption The sparse reward plus equality following constraint has a feasible solution in all test scenarios.
    The CMDP in Eq. (3) assumes there exists a policy that keeps E[C_F] exactly at delta_F while also driving reward; in dense or rushing crowds this may be infeasible, yet no infeasibility handling is discussed.
  • standard math CMDP formulation with GAE and PPO clipping remains valid for the multi-cost objective.
    Standard result from references [8], [9], [25], and [27]; the paper applies it as-is without modification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Navigating the Proximity-Safety Balance: Constraint Decomposition for Human Following in Pedestrian Crowds." pith.science (2026). https://pith.science/paper/RFG6LOK7

@misc{pith2026260810056,
  author       = {Pith},
  title        = {Pith review of: Navigating the Proximity-Safety Balance: Constraint Decomposition for Human Following in Pedestrian Crowds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RFG6LOK7}},
  note         = {Machine review of arXiv:2608.10056}
}
read the original abstract

Following a target human in crowded environments involves an inherent conflict between staying close to the target and navigating safely among surrounding pedestrians and obstacles. This conflict becomes more severe in dense scenarios, where aggressive following risks collisions and conservative margins lead to target loss, especially when pedestrian behaviors are unfamiliar or unpredictable. Existing reinforcement learning (RL) methods typically encode these competing objectives into a single dense reward, but the resulting proximity-safety balance is implicit and difficult to adjust across conditions. To address this, we decompose the human-following task into a sparse task reward and independent cost constraints within a multi-constraint RL formulation, where each constraint is managed through cost thresholds with direct behavioral meaning rather than implicit reward weight ratios, allowing explicit and tunable control over the trade-off. We further quantify the prediction uncertainty of human motions and integrate these estimates into the RL costs to enhance safety under unpredictable conditions. Extensive experiments across both in-distribution and out-of-distribution settings demonstrate that our method achieves an effective proximity-safety balance compared to baselines. Real-robot deployment further validates the feasibility of our method in real-world scenarios. More details are available on our project page: https://nav-ps-balance.github.io/.

Figures

Figures reproduced from arXiv: 2608.10056 by the authors.

Figure 1
Figure 1. An overview diagram of our method. Components related to static [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The interaction among the critics and the generation of the action [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visualization of test results. Regular pedestrians are shown in blue, the target pedestrian is represented in red, and the robot is shown in orange. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 19 canonical work pages

  1. [20]

    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,” inConference on Robot Learning (CoRL), 2025

    work page 2025

  2. [22]

    Sonic: Safe social navigation with adaptive conformal inference and constrained reinforcement learning,

    J. Yao, X. Zhang, Y . Xia, Z. Wang, A. K. Roy-Chowdhury, and J. Li, “Sonic: Safe social navigation with adaptive conformal inference and constrained reinforcement learning,”arXiv preprint arXiv:2407.17460, 2024

    arXiv 2024

  3. [1]

    Exploring the role of human-following robots in supporting the mobility and wellbeing of older people,

    S. Li, K. Milligan, P. Blythe, Y . Zhang, S. Edwards, N. Palmarini, L. Corner, Y . Ji, F. Zhang, and A. Namdeo, “Exploring the role of human-following robots in supporting the mobility and wellbeing of older people,”Scientific reports, vol. 13, no. 1, p. 6512, 2023

    work page 2023

  4. [2]

    Human-following and-guiding in crowded environments using se- mantic deep-reinforcement-learning for mobile service robots,

    L. K ¨astner, B. Fatloun, Z. Shen, D. Gawrisch, and J. Lambrecht, “Human-following and-guiding in crowded environments using se- mantic deep-reinforcement-learning for mobile service robots,” in 2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 833–839

    work page 2022

  5. [3]

    An mcts-drl based obstacle and occlusion avoidance methodology in robotic follow- ahead applications,

    S. Leisiazar, E. J. Park, A. Lim, and M. Chen, “An mcts-drl based obstacle and occlusion avoidance methodology in robotic follow- ahead applications,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2023, pp. 221–228

    work page 2023

  6. [4]

    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

    2021

  7. [5]

    Conformal inference for online predic- tion with arbitrary distribution shifts,

    I. Gibbs and E. J. Cand `es, “Conformal inference for online predic- tion with arbitrary distribution shifts,”Journal of Machine Learning Research, vol. 25, no. 162, pp. 1–36, 2024

    work page 2024

  8. [6]

    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

Show all 37 references
  1. [7]

    Navformer: A transformer ar- chitecture for robot target-driven navigation in unknown and dynamic environments,

    H. Wang, A. H. Tan, and G. Nejat, “Navformer: A transformer ar- chitecture for robot target-driven navigation in unknown and dynamic environments,”IEEE Robotics and Automation Letters, vol. 9, no. 8, pp. 6808–6815, 2024

    2024

  2. [8]

    Omnisafe: An infrastructure for accelerating safe reinforcement learning research,

    J. Ji, J. Zhou, B. Zhang, J. Dai, X. Pan, R. Sun, W. Huang, Y . Geng, M. Liu, and Y . Yang, “Omnisafe: An infrastructure for accelerating safe reinforcement learning research,”Journal of Machine Learning Research, vol. 25, no. 285, pp. 1–6, 2024

    2024

  3. [9]

    Benchmarking safe exploration in deep reinforcement learning,

    A. Ray, J. Achiam, and D. Amodei, “Benchmarking safe exploration in deep reinforcement learning,”arXiv preprint arXiv:1910.01708, 2019

    1910 arXiv

  4. [10]

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

    C. Chen, Y . Liu, S. Kreiss, and A. Alahi, “Crowd-robot interaction: Crowd-aware robot navigation with attention-based deep reinforce- ment learning,” in2019 International Conference on Robotics and Automation (ICRA). IEEE, 2019, pp. 6015–6022

    2019

  5. [11]

    Person-following by autonomous robots: A categorical overview,

    M. J. Islam, J. Hong, and J. Sattar, “Person-following by autonomous robots: A categorical overview,”The International Journal of Robotics Research, vol. 38, no. 14, pp. 1581–1618, 2019

    2019

  6. [12]

    Human following and guidance by autonomous mobile robots: A comprehensive review,

    A. Eirale, M. Martini, and M. Chiaberge, “Human following and guidance by autonomous mobile robots: A comprehensive review,” IEEE Access, 2025

    2025

  7. [13]

    Obstacle-avoidant leader following with a quadruped robot,

    C. Scheidemann, L. Werner, V . Reijgwart, A. Cramariuc, J. Chomarat, J.-R. Chiu, R. Siegwart, and M. Hutter, “Obstacle-avoidant leader following with a quadruped robot,” in2025 IEEE International Con- ference on Robotics and Automation (ICRA). IEEE, 2025, pp. 1407– 1413

    2025

  8. [14]

    Robot person following under partial occlusion,

    H. Ye, J. Zhao, Y . Pan, W. Chen, L. He, and H. Zhang, “Robot person following under partial occlusion,” in2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023, pp. 7591–7597

    2023

  9. [15]

    Trackvla: Embodied visual tracking in the wild,

    S. Wang, J. Zhang, M. Li, J. Liu, A. Li, K. Wu, F. Zhong, J. Yu, Z. Zhang, and H. Wang, “Trackvla: Embodied visual tracking in the wild,” inConference on Robot Learning. PMLR, 2025, pp. 4139– 4164

    2025

  10. [16]

    Safe and robust human follow- ing for mobile robots based on self-avoidance mpc in crowded corridor scenarios,

    Y . Song, Q. Zhang, Z. Hu, and J. Liu, “Safe and robust human follow- ing for mobile robots based on self-avoidance mpc in crowded corridor scenarios,” in2023 IEEE International Conference on Robotics and Biomimetics (ROBIO). IEEE, 2023, pp. 1–6

    2023

  11. [17]

    Adap-rpf: Adaptive trajectory sampling for robot person following in dynamic crowded environments,

    W. Situ, H. Ye, J. Peng, Y . Zhan, and H. Zhang, “Adap-rpf: Adaptive trajectory sampling for robot person following in dynamic crowded environments,”arXiv preprint arXiv:2510.11308, 2025

    2025

  12. [18]

    Not only rewards but also constraints: Applications on legged robot locomotion,

    Y . Kim, H. Oh, J. Lee, J. Choi, G. Ji, M. Jung, D. Youm, and J. Hwangbo, “Not only rewards but also constraints: Applications on legged robot locomotion,”IEEE Transactions on Robotics, vol. 40, pp. 2984–3003, 2024

    2024

  13. [19]

    Evaluation of constrained reinforcement learning algorithms for legged locomotion,

    J. Lee, L. Schroth, V . Klemm, M. Bjelonic, A. Reske, and M. Hut- ter, “Evaluation of constrained reinforcement learning algorithms for legged locomotion,”arXiv preprint arXiv:2309.15430, 2023

    2023 arXiv

  14. [21]

    Safe learning in robotics: From learning-based control to safe reinforcement learning,

    L. Brunke, M. Greeff, A. W. Hall, Z. Yuan, S. Zhou, J. Panerati, and A. P. Schoellig, “Safe learning in robotics: From learning-based control to safe reinforcement learning,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 5, no. 1, pp. 411–444, 2022

    2022

  15. [23]

    Constrained policy op- timization,

    J. Achiam, D. Held, A. Tamar, and P. Abbeel, “Constrained policy op- timization,” inInternational conference on machine learning. PMLR, 2017, pp. 22–31

    2017

  16. [24]

    Projection- based constrained policy optimization,

    T.-Y . Yang, J. Rosca, K. Narasimhan, and P. J. Ramadge, “Projection- based constrained policy optimization,” inInternational Conference on Learning Representations, 2020

    2020

  17. [25]

    Altman,Constrained Markov decision processes

    E. Altman,Constrained Markov decision processes. Routledge, 2021

    2021

  18. [26]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems, vol. 30, 2017

    2017

  19. [27]

    High- dimensional continuous control using generalized advantage estima- tion,

    J. Schulman, P. Moritz, S. Levine, M. Jordan, and P. Abbeel, “High- dimensional continuous control using generalized advantage estima- tion,” inInternational Conference on Learning Representations, 2016

    2016

  20. [28]

    Wayformer: Motion forecasting via simple & efficient atten- tion networks,

    N. Nayakanti, R. Al-Rfou, A. Zhou, K. Goel, K. S. Refaat, and B. Sapp, “Wayformer: Motion forecasting via simple & efficient atten- tion networks,” in2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023, pp. 2980–2987

    2023

  21. [29]

    Height: Heterogeneous interac- tion graph transformer for robot navigation in crowded and constrained environments,

    S. Liu, H. Xia, F. C. Pouria, K. Hong, N. Chakraborty, Z. Hu, J. Biswas, and K. Driggs-Campbell, “Height: Heterogeneous interac- tion graph transformer for robot navigation in crowded and constrained environments,”IEEE Transactions on Automation Science and Engi- neering, vol....

    2026

  22. [30]

    Proximal policy optimization algorithms,

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,”arXiv preprint arXiv:1707.06347, 2017

    2017 arXiv

  23. [31]

    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

  24. [32]

    Efficient motion planning based on kinodynamic model for quadruped robots following persons in confined spaces,

    Z. Zhang, J. Yan, X. Kong, G. Zhai, and Y . Liu, “Efficient motion planning based on kinodynamic model for quadruped robots following persons in confined spaces,”IEEE/ASME Transactions on Mechatron- ics, vol. 26, no. 4, pp. 1997–2006, 2021

    1997

  25. [33]

    Path planning for autonomous vehicles in unknown semi-structured environments,

    D. Dolgov, S. Thrun, M. Montemerlo, and J. Diebel, “Path planning for autonomous vehicles in unknown semi-structured environments,” The international journal of robotics research, vol. 29, no. 5, pp. 485– 501, 2010

    2010

  26. [34]

    Interaction-aware conformal prediction for crowd naviga- tion,

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

    2025 arXiv

  27. [35]

    What the constant velocity model can teach us about pedestrian motion prediction,

    C. Sch ¨oller, V . Aravantinos, F. Lay, and A. Knoll, “What the constant velocity model can teach us about pedestrian motion prediction,”IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 1696–1703, 2020

    2020

  28. [36]

    Dr-spaam: A spatial-attention and auto-regressive model for person detection in 2d range data,

    D. Jia, A. Hermans, and B. Leibe, “Dr-spaam: A spatial-attention and auto-regressive model for person detection in 2d range data,” in2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2020, pp. 10 270–10 277

    2020

  29. [37]

    Simple online and realtime tracking,

    A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, “Simple online and realtime tracking,” in2016 IEEE international conference on image processing (ICIP). IEEE, 2016, pp. 3464–3468

    2016

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.