Pith. sign in

REVIEW 5 major objections 5 minor 31 references

Dexterous Safe Control for Humanoids in Cluttered Environments via Projected Safe Set Algorithm

T0 review · 5 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

Pith's one-line read Projected Safe Set Algorithm (p-SSA) splits feasibility projection from task tracking, so a humanoid's multi-constraint safety QP is always feasible and needs no parameter tuning.

desk verdict Deserves peer review: the two-phase slack-projection idea is clean and the problem setting is real, but the paper overstates a surrogate-objective result as a safety guarantee. read the letter →

arxiv 2502.02858 v1 pith:QEVW5G5W submitted 2025-02-05 cs.RO

classification cs.RO MSC 93C8590C20
keywords dexteroussafetysafesetalgorithmcontrolbarrierfunctionshumanoidrobotscollisionavoidancequadraticprogramminginfeasibleconstraintsteleoperation
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 tackles dexterous safety for humanoids: avoiding external and self-collisions with limb-level geometry in cluttered environments. Such tasks produce hundreds of safety constraints, and the quadratic programs used by standard safe controllers often become infeasible. The authors propose the Projected Safe Set Algorithm (p-SSA), which first solves a slack-minimization phase to project the infeasible constraint set onto the nearest feasible set, then solves the tracking QP with that fixed relaxation. They claim p-SSA always operates within the maximal feasible region, is completely tuning-free, and outperforms the naive SSA and relaxed SSA baselines on all eight simulated tasks, with a real Unitree G1 teleoperation demonstration.

What carries the argument

The central object is the projection of an infeasible safe-control constraint set onto the nearest feasible set, measured by the $p$-norm of total relaxation. Phase I (12) minimizes that relaxation, choosing slack variables $s^*$; Phase II (13) then solves the original tracking objective $\|u - u_{\text{ref}}\|^2_{2,Q}$ subject to the safety constraints relaxed by exactly $s^*$. Because $s^*$ is feasible for (12), (13) is guaranteed to have a solution. The decoupling removes the trade-off between safety and performance that plagues weighted slack regularization, which is what makes p-SSA parameter-free.

What would settle it

Run p-SSA on a humanoid arm with a deliberately degraded velocity tracker (e.g., 200 ms latency or 30% gain error) inside a narrow cavity, while recording actual collisions; if the QP remains feasible but the robot still hits the obstacle, the accuracy assumption is refuted. Equivalently, construct a two-plane example where the minimal-slack projection moves the end-effector parallel to the planes while the true distance still decreases, showing that minimal constraint slack is not minimal physical violation.

Watch

Extended reading notes

Core claim

The paper's central claim is that p-SSA always operates within the maximal feasible region indicated by the given constraint set while being totally tuning-free, and that it achieves top performance across various task settings without parameter tuning. Concretely, the two-stage procedure—minimize the slack $s$ in (12), then solve the tracking QP with the optimal $s^*$ in (13)—guarantees that the second QP is feasible by construction. In experiments p-SSA yields lower control-constraint violation and distance-safety violation scores than naive SSA and r-SSA on all eight simulated dexterous safety tasks, and it prevents collisions in a real Unitree G1 teleoperation test where the operator deliberately makes risky motions.

Load-bearing premise

The load-bearing premise is that the robot's low-level controller tracks commanded joint velocities accurately enough that the first-order model $\dot{x}=u$ matches the real robot, so the linearized safety constraints describe the true evolution of the safety index.

Editorial extensions

If this is right

  • p-SSA can be deployed directly on high-dimensional humanoid models with hundreds of collision-avoidance constraints without per-task tuning.
  • When a collision is physically inevitable, p-SSA provides a principled minimal-violation response instead of passing through an arbitrary reference.
  • The same projection idea extends to any energy-function safe controller that yields a control-affine QP, including CBF-based filters.
  • Joint-space velocity commands computed by p-SSA can act as a safety filter for teleoperation interfaces, allowing operators to issue unsafe goals while the robot remains safe.
  • The guarantee that the second QP is feasible holds even under dynamic obstacles, since feasibility comes from the projection, not from the obstacle distribution.

Reading between the lines

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

  • We infer that the slack amount $s^*$ solved in Phase I can serve as an online indicator of how close the current situation is to local infeasibility, something the paper does not exploit.
  • The paper sets the slack weighting to identity and leaves smart balancing to future work; an obvious test is whether distance- or torque-aware weights push p-SSA's Pareto frontier further.
  • Because the projection step only needs the constraint set, the same Phase-I/Phase-II split could be applied to CBF-based multi-constraint QPs, which face the same infeasibility problem.
  • The linearization in the safety constraints means the 'minimal violation' is minimal in constraint space, not necessarily in physical distance; we infer this gap would grow under fast dynamics and imperfect tracking.
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

5 major / 5 minor

Summary. The paper addresses dexterous safe control for humanoid robots in cluttered environments, where a multi-constraint safe control QP with hundreds of collision-avoidance constraints frequently becomes infeasible. The authors propose r-SSA, which adds weighted slack variables to relax the safety constraints, and p-SSA, a two-phase method that first computes a minimal slack vector to render the constraints feasible and then solves a tracking QP with that slack fixed, guaranteeing feasibility of the second QP by construction. The paper reports simulation results on eight tasks with a Unitree G1 model and a real hardware teleoperation demonstration, claiming minimal safety violations, zero parameter tuning, and top performance relative to baselines.

Significance. The Phase I / Phase II decoupling is a clean idea with practical value: it avoids the safety-performance weight tuning of r-SSA and guarantees feasibility of the tracking QP whenever the projection step is solved, which is nontrivial for high-dimensional multi-constraint problems. The real-robot demonstration is also a useful step toward deployable dexterous safety. However, the headline claims are not fully supported. The 'minimal safety violations' assertion concerns a slack variable in a linearized derivative constraint, not the distance-based safety specification, and the reported advantage is partly built into the evaluation metric. The velocity-tracking assumption underlying the hardware experiments is not validated. With a corrected evaluation and a more careful statement of what is minimized, the contribution would be solid.

major comments (5)
  1. [Section V-D, Eq. (15); Section IV-B, Eq. (12)] The 'minimal safety violations' claim is not supported by the evidence as presented. In Eq. (15), the C score for p-SSA uses s_i,t from the Phase I solution of (12), but the executed control is the Phase II solution of (13); the realized control-constraint slack of that control, max(dot-phi_i(x_t,u_t)+eta_i,0), is generally different from the Phase I slack, so the reported C advantage is partly an artifact of the projection rather than of the actual behavior. More importantly, the Phase I objective is a norm of the slack in the linearized derivative constraint (12b), not a measure of distance-based safety violation; no theorem or experiment connects this slack to the S metric in (16). Section VI explicitly concedes that once relaxation is active, no safety guarantee holds. The abstract and Section IV-B should either rephrase the claim to 'minimal violation of the linearized safety constraints' or provide a concrete link, such as a bound S <= F(||s||) for the first-order model.
  2. [Section V-A and V-G] The safety constraints are derived from the first-order model x_dot = u, with the text stating 'assuming an accurate velocity tracker at the downstream.' In the real Unitree G1 experiment, no tracking-error measurements or robustness margins are reported. If the low-level velocity tracker is imperfect, the inequality (9b) does not describe the actual evolution of the safety index, and a feasible p-SSA solution can still permit collisions. The paper should either quantify the tracking error on the real robot, add a margin to dmin to absorb it, or explicitly limit the hardware claim to an empirical demonstration rather than a safety guarantee.
  3. [Section IV-B and abstract] The claim of being 'totally tuning-free' / 'zero parameter tuning' is overstated. p-SSA still requires choosing eta, the safety margins dmin,env and dmin,self, the norm order p, and Qpssa_s (set to identity in the paper). What is tuning-free is the relative weight between the tracking objective and the slack penalty, compared with r-SSA's Qrssa_s. Please rephrase to 'without tuning the safety-performance trade-off weight'.
  4. [Section IV-B, Eq. (12)] The Phase I optimization is written as minimizing over s only, while u appears in the constraints (12b)-(12c). As written, u is not a decision variable and the problem is not well-posed. To guarantee the Phase II feasibility argument, Phase I must be understood as an optimization over (u,s), or at least as an existential projection over u. Please correct the notation and specify the metric with respect to which the constraint set is projected.
  5. [Section V-E and V-F] The headline comparisons are based on a single 2000-step trajectory per task, with no multiple seeds, error bars, or statistical significance tests. Since the dynamic-obstacle tasks use Brownian motion, run-to-run variance is non-negligible, and the Pareto-front plots in Figures 7 and 8 represent point estimates. The claim that p-SSA 'automatically secures the sweet spot' on the Pareto fronts needs at least a few seeds per configuration or a statistical comparison.
minor comments (5)
  1. [Abstract] Typo: 'sprase' should be 'sparse'.
  2. [Figure 9 caption] Typo: 'simluated' should be 'simulated' and 'teleopration' should be 'teleoperation'.
  3. [Section V-E-a] The statement 'the naive SSA essentially ignores the obstacles if the QP is infeasible' is imprecise; SSA actually passes uref unmodified when the QP is infeasible, which may or may not avoid obstacles. Consider rewording.
  4. [Section V-D, Eq. (15)] The C and S metrics are undefined (0/0) when there are no infeasible steps or no distance violations, respectively. Please state the convention used in the plots for these cases.
  5. [Section IX-C, Figure 10] The caption states that p-SSA 'consistently bounds phi below zero,' but the text acknowledges a tolerance due to discretization. Please quantify this tolerance and the conditions under which it is bounded.

Circularity Check

1 steps flagged · score 2.0 of 10

One metric-level circularity: the C score is computed from the same slack variables p-SSA minimizes, but the central safety claim is independently supported by the S metric and hardware demo.

  1. self definitional [Section V-D, Eq. (15); Section IV-B, Eq. (12)]
    "C: control constraint satisfaction score, given by C = Σ_t I(s_t > 0) exp(−s_t^2/0.2) / Σ_t I(s_t > 0) (15) ... For p-SSA, s_{i,t} is the solution to (12)."

    Phase I of p-SSA, Eq. (12a), minimizes (1/p)||s||_p^p, and Eq. (15) is a decreasing, exponentiated function of the total slack s_t = Σ_i s_{i,t}; for p-SSA, the s values entering Eq. (15) are exactly the optimal s* produced by Eq. (12). Hence p-SSA's high C score is guaranteed by construction: it is the same quantity minimized in Phase I. Consequently, the statement 'p-SSA guarantees to operate with minimal safety violations thanks to Phase I' is only definitionally true for the slack surrogate, not for distance-based safety. The paper's independent S metric (Eq. 16) measures actual margin invasion and does not reduce to the p-SSA objective, so this is a metric-level tautology rather than a full collapse of the central claim.

full rationale

The derivation chain is otherwise self-contained. The p-SSA feasibility guarantee is constructive: Phase I finds any feasible slack s*, which is then fixed in Phase II, so Eq. (13) is feasible by construction; no parameter is fitted to the evaluation data and then reported as a prediction. No load-bearing self-citation or imported uniqueness theorem was found; self-citations are to prior SSA/SIS work and are not used to justify the core projection argument. The only self-referential element is the C metric, which is definitionally aligned with the p-SSA objective. This does not invalidate the central contribution because the paper also reports the independent distance-based S metric and a real-robot teleoperation demonstration. Section VI candidly states that 'as long as relaxation is needed, the safety violation cannot be bounded, preventing any safety guarantee to hold,' which appropriately limits the strength of the safety claim and further supports the view that the paper is not systematically circular.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The p-SSA design depends on a handful of hand-set constants (eta, dmin, norm order, Q, Qs) and on several domain assumptions about model fidelity and sensing. There are no physically invented entities. The main load-bearing conceptual assumption is that minimal slack in the control-constraint space is a faithful proxy for minimal real-world safety violation; the paper itself notes in Section VI that once relaxation is active, safety violations are unbounded.

free parameters (7)
  • eta (safety index decay rate) = 0.5 for all constraints
    Chosen in Section V-C as the required decrease rate for each safety index; it affects conservatism and QP feasibility and is a hand-set parameter.
  • dmin_env (external safety margin) = 0.05 m
    Minimum distance to external obstacles in Section V-B; task-specific and chosen by hand.
  • dmin_self (self-collision margin) = 0.01 m
    Minimum distance between self-collision pairs in Section V-B; task-specific and chosen by hand.
  • p (norm order for slack) = 2
    Norm order in Eqs. (10) and (12); chosen without comparison to other p values.
  • Q (tracking objective weight) = Identity
    Weight matrix in the tracking objective (13); set to identity in all experiments.
  • Q^pssa_s (p-SSA slack weight) = Identity
    Diagonal weight for slack in Phase I; the paper sets it to identity and defers smart weighting to future work.
  • Q^rssa_s (r-SSA slack weight) = 10I for main comparison; ablated over a grid of a*10^b values
    Weight balancing performance and safety in r-SSA; the ablation study shows the trade-off that p-SSA avoids.
assumptions (5)
  • standard math Dynamics are control-affine with bounded control (Eq. (1)).
    Invoked in Section III-A to set up the QP formulations (9), (10), (12), and (13).
  • domain assumption First-order integrator model x_dot = u with an accurate velocity tracker.
    Stated in Section V-A; the derived safety constraints in Appendix IX-B rely on x_dot = u, so the real robot must track joint velocity commands closely for the constraints to hold on hardware.
  • domain assumption Collision safety is fully captured by pairwise distance lower bounds between sphere models of robot links and obstacles or body pairs.
    Used in Section V-C and Appendix IX-B; the method treats each pair independently with a first-order safety index and does not account for other collision geometries or higher-order effects.
  • ad hoc to paper Relaxing the control constraints minimally (small slack s) is a valid proxy for minimal physical safety violation.
    This design principle underlies Phase I of p-SSA (Eq. (12)) and the claim of minimal safety violations; it is not proven and is partially contradicted by the limitations section, which states that no safety guarantee holds once relaxation is active.
  • domain assumption Obstacle positions and robot states are known and tracked (via Apple Vision Pro for hardware).
    The safety filter computes distances and Jacobians from current state and obstacle locations; sensing errors would degrade the computed constraints.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dexterous Safe Control for Humanoids in Cluttered Environments via Projected Safe Set Algorithm." pith.science (2026). https://pith.science/paper/QEVW5G5W

@misc{pith2026250202858,
  author       = {Pith},
  title        = {Pith review of: Dexterous Safe Control for Humanoids in Cluttered Environments via Projected Safe Set Algorithm},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QEVW5G5W}},
  note         = {Machine review of arXiv:2502.02858}
}
read the original abstract

It is critical to ensure safety for humanoid robots in real-world applications without compromising performance. In this paper, we consider the problem of dexterous safety, featuring limb-level geometry constraints for avoiding both external and self-collisions in cluttered environments. Compared to safety with simplified bounding geometries in sprase environments, dexterous safety produces numerous constraints which often lead to infeasible constraint sets when solving for safe robot control. To address this issue, we propose Projected Safe Set Algorithm (p-SSA), an extension of classical safe control algorithms to multi-constraint cases. p-SSA relaxes conflicting constraints in a principled manner, minimizing safety violations to guarantee feasible robot control. We verify our approach in simulation and on a real Unitree G1 humanoid robot performing complex collision avoidance tasks. Results show that p-SSA enables the humanoid to operate robustly in challenging situations with minimal safety violations and directly generalizes to various tasks with zero parameter tuning.

Figures

Figures reproduced from arXiv: 2502.02858 by the authors.

Figure 1
Figure 1. Application of dexterous safe control for humanoids in cluttered environments. (a) A safe teleoperation task where the Unitree G1 humanoid mirrors [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Possible scenarios where (6) can be infeasible. The humanoid should [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Unitree G1 humanoid robot in MuJoCo simulation performing safe [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Comparison of safe control methods in G1FixedBase [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Performance comparison under G1WholeBody configuration [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Performance comparison under G1FixedBase configuration. Since we disable the physical collision for simulated obstacles, the humanoid bodies can quickly swing through the obstacles and enter empty spaces. That makes QPs solved by naive SSA to have less active constrain…
Figure 7
Figure 7. Figure 7: Ablation study on r-SSA and p-SSA with G1WholeBody. Pareto fronts are plotted for r-SSA. r-SSA points are less transparent for larger [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Ablation study on r-SSA and p-SSA with G1FixedBody. Pareto fronts are plotted for r-SSA. r-SSA points are less transparent for larger [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Safe teleopration with simluated Unitree G1 humanoid. The humanoid tracks wrist position goals (green) sent by the tele-operator while avoiding [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Change of ϕ that describes safety between the left hand and the nearest obstacle in fig. 4. We can see that p-SSA consistently bounds ϕ below zero with some tolerance due to discretization. r-SSA allows more violations when handling infeasible QP. Naive SSA (No Slack)…
Figure 11
Figure 11. Figure 11: Change of left elbow joint position in fig. 4. [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 12
Figure 12. Figure 12: Change of left shoulder roll joint position in fig. 4. [PITH_FULL_IMAGE:figures/full_fig_p015_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 28 canonical work pages

  1. [1]

    Control barrier function based quadratic programs with application to adaptive cruise control

    Aaron D Ames, Jessy W Grizzle, and Paulo Tabuada. Control barrier function based quadratic programs with application to adaptive cruise control. In 53rd IEEE Conference on Decision and Control , pages 6271–6278. IEEE, 2014

  2. [2]

    Control barrier function based quadratic programs for safety critical systems

    Aaron D Ames, Xiangru Xu, Jessy W Grizzle, and Paulo Tabuada. Control barrier function based quadratic programs for safety critical systems. IEEE Transactions on Automatic Control , 62(8):3861–3876, 2016

  3. [3]

    Compositions of multiple control barrier functions under input constraints

    Joseph Breeden and Dimitra Panagou. Compositions of multiple control barrier functions under input constraints. In 2023 American Control Conference (ACC) , pages 3688–3695. IEEE, 2023

  4. [4]

    Safe and sample- efficient reinforcement learning for clustered dynamic environments

    Hongyi Chen and Changliu Liu. Safe and sample- efficient reinforcement learning for clustered dynamic environments. IEEE Control Systems Letters , 6:1928– 1933, 2021

  5. [5]

    Safety index synthesis with state-dependent control space

    Rui Chen, Weiye Zhao, and Changliu Liu. Safety index synthesis with state-dependent control space. arXiv preprint, 2023

  6. [6]

    Real-time safety index adaptation for parameter-varying systems via determinant gradient ascend

    Rui Chen, Weiye Zhao, Ruixuan Liu, Weiyang Zhang, and Changliu Liu. Real-time safety index adaptation for parameter-varying systems via determinant gradient ascend. arXiv preprint, 2023

  7. [7]

    Robust control barrier– value functions for safety-critical control

    Jason J Choi, Donggun Lee, Koushil Sreenath, Claire J Tomlin, and Sylvia L Herbert. Robust control barrier– value functions for safety-critical control. In IEEE CDC, pages 6814–6821, 2021

  8. [8]

    Constraint-Guided Online Data Selection for Scalable Data-Driven Safety Filters in Uncertain Robotic Systems

    Jason J Choi, Fernando Castaneda, Wonsuhk Jung, Bike Zhang, Claire J Tomlin, and Koushil Sreenath. Constraint-guided online data selection for scalable data- driven safety filters in uncertain robotic systems. arXiv preprint arXiv:2311.13824, 2023

Show all 31 references
  1. [9]

    Safe nonlinear control using robust neural lyapunov- barrier functions

    Charles Dawson, Zengyi Qin, Sicun Gao, and Chuchu Fan. Safe nonlinear control using robust neural lyapunov- barrier functions. In Conference on Robot Learning , pages 1724–1735. PMLR, 2022

  2. [10]

    Robust task-space quadratic programming for kinematic-controlled robots

    Mohamed Djeha, Pierre Gergondet, and Abderrahmane Kheddar. Robust task-space quadratic programming for kinematic-controlled robots. IEEE Transactions on Robotics, 2023

  3. [11]

    Hybrid nonsmooth barrier functions with applications to provably safe and composable collision avoidance for robotic systems

    Paul Glotfelter, Ian Buckley, and Magnus Egerstedt. Hybrid nonsmooth barrier functions with applications to provably safe and composable collision avoidance for robotic systems. IEEE Robotics and Automation Letters , 4(2):1303–1310, 2019

  4. [12]

    Agile but safe: Learn- ing collision-free high-speed legged locomotion

    Tairan He, Chong Zhang, Wenli Xiao, Guanqi He, Changliu Liu, and Guanya Shi. Agile but safe: Learn- ing collision-free high-speed legged locomotion. arXiv preprint arXiv:2401.17583, 2024

  5. [13]

    Humanoid self-collision avoidance using whole-body control with control barrier functions

    Charles Khazoom, Daniel Gonzalez-Diaz, Yanran Ding, and Sangbae Kim. Humanoid self-collision avoidance using whole-body control with control barrier functions. In 2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) , pages 558–565. IEEE, 2022

  6. [14]

    Real-time collision avoidance algorithm on industrial manipulators

    Hsien-Chung Lin, Changliu Liu, Yongxiang Fan, and Masayoshi Tomizuka. Real-time collision avoidance algorithm on industrial manipulators. In IEEE CCTA , pages 1294–1299. IEEE, 2017

  7. [15]

    Control in a safe set: Addressing safety in human-robot interac- tions

    Changliu Liu and Masayoshi Tomizuka. Control in a safe set: Addressing safety in human-robot interac- tions. In Dynamic Systems and Control Conference , vol- ume 46209. American Society of Mechanical Engineers, 2014

  8. [16]

    Safe interac- tive industrial robots using jerk-based safe set algorithm

    Ruixuan Liu, Rui Chen, and Changliu Liu. Safe interac- tive industrial robots using jerk-based safe set algorithm. arXiv preprint arXiv:2204.03038 , 2022

  9. [17]

    Proactive human-robot co-assembly: Leveraging human intention prediction and robust safe control

    Ruixuan Liu, Rui Chen, Abulikemu Abuduweili, and Changliu Liu. Proactive human-robot co-assembly: Leveraging human intention prediction and robust safe control. In 2023 IEEE Conference on Control Technology and Applications (CCTA) , pages 339–345. IEEE, 2023

  10. [18]

    Safe control under input saturation with neural control barrier func- tions

    Simin Liu, John Dolan, and Changliu Liu. Safe control under input saturation with neural control barrier func- tions. In CoRL, 2022

  11. [19]

    Model-free safety-critical control for robotic systems

    Tamas G Molnar, Ryan K Cosner, Andrew W Singletary, Wyatt Ubellacker, and Aaron D Ames. Model-free safety-critical control for robotic systems. IEEE robotics and automation letters , 7(2):944–951, 2021

  12. [20]

    3d dynamic walking on stepping stones with control barrier functions

    Quan Nguyen, Ayonga Hereid, Jessy W Grizzle, Aaron D Ames, and Koushil Sreenath. 3d dynamic walking on stepping stones with control barrier functions. In 2016 IEEE 55th Conference on Decision and Control (CDC) , pages 827–834. IEEE, 2016

  13. [21]

    Multi- modal safe control for human-robot interaction

    Ravi Pandya, Tianhao Wei, and Changliu Liu. Multi- modal safe control for human-robot interaction. In 2024 American Control Conference (ACC) , pages 2672–2678. IEEE, 2024

  14. [22]

    Onboard safety guarantees for racing drones: High-speed geofencing with control barrier func- tions

    Andrew Singletary, Aiden Swann, Yuxiao Chen, and Aaron D Ames. Onboard safety guarantees for racing drones: High-speed geofencing with control barrier func- tions. IEEE Robotics and Automation Letters, 7(2):2897– 2904, 2022

  15. [23]

    Learning for safety-critical control with control barrier functions

    Andrew Taylor, Andrew Singletary, Yisong Yue, and Aaron Ames. Learning for safety-critical control with control barrier functions. In Learning for Dynamics and Control, pages 708–717. PMLR, 2020

  16. [24]

    Adaptive safety with control barrier functions

    Andrew J Taylor and Aaron D Ames. Adaptive safety with control barrier functions. In 2020 American Control Conference (ACC), pages 1399–1405. IEEE, 2020

  17. [25]

    High-order control barrier functions-based optimization control for time-varying nonlinear systems with full-state constraints: A dynamic sub-safe set approach

    Haijing Wang, Jinzhu Peng, Juanjuan Xu, Fangfang Zhang, and Yaonan Wang. High-order control barrier functions-based optimization control for time-varying nonlinear systems with full-state constraints: A dynamic sub-safe set approach. International Journal of Robust and Nonline...

  18. [26]

    Safe control algorithms using energy functions: A unified framework, bench- mark, and new directions

    Tianhao Wei and Changliu Liu. Safe control algorithms using energy functions: A unified framework, bench- mark, and new directions. In Conference on Decision and Control, 2019

  19. [27]

    Control barrier functions for systems with high relative degree

    Wei Xiao and Calin Belta. Control barrier functions for systems with high relative degree. In CDC, pages 474–

  20. [28]

    Safe control of quadruped in varying dynamics via safety index adaptation

    Kai S Yun, Rui Chen, Chase Dunaway, John M Dolan, and Changliu Liu. Safe control of quadruped in varying dynamics via safety index adaptation. arXiv preprint arXiv:2409.09882, 2024

  21. [29]

    Model-free safe control for zero-violation reinforcement learning

    Weiye Zhao, Tairan He, and Changliu Liu. Model-free safe control for zero-violation reinforcement learning. In Conference on Robot Learning , 2021

  22. [30]

    Safety index synthesis via sum-of-squares programming

    Weiye Zhao, Tairan He, Tianhao Wei, Simin Liu, and Changliu Liu. Safety index synthesis via sum-of-squares programming. In ACC, pages 732–737. IEEE, 2023

  23. [31]

    Model free safe control for reinforcement learning in a clus- tered dynamic environment

    Guiliang Zheng, Minhao Yang, and Yuxuan Wu. Model free safe control for reinforcement learning in a clus- tered dynamic environment. In 2022 5th Interna- tional Conference on Mechatronics, Robotics and Au- tomation (ICMRA) , pages 19–26, 2022. doi: 10.1109/ ICMRA56206.2022.101...

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.