pith. sign in

arxiv: 2505.00306 · v5 · submitted 2025-05-01 · 💻 cs.RO

J-PARSE: Jacobian-based Projection Algorithm for Resolving Singularities Effectively in Inverse Kinematic Control of Serial Manipulators

Shivani Guptasarma , Matthew Strong , Honghao Zhen , Monroe Kennedy III This is my paper

Pith reviewed 2026-05-22 18:21 UTC · model grok-4.3

classification 💻 cs.RO
keywords inverse kinematicskinematic singularitiesserial manipulatorsJacobian matrixmanipulability ellipsoidvelocity controlrobot motion planning
0
0 comments X

The pith

J-PARSE resolves singularities in inverse kinematics by constructing a Safety Jacobian and projecting velocities onto safe directions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents J-PARSE as a method for smooth first-order inverse kinematic control of serial manipulators near singularities. It builds a substitute Safety Jacobian that preserves a minimum aspect ratio in the manipulability ellipsoid, then splits the desired end-effector velocity into non-singular and singular components and scales the singular part before applying the right inverse of the modified Jacobian. This construction is shown to produce continuous velocity commands that allow safe passage through rank-deficient configurations. In the absence of joint limits or collisions the method delivers asymptotic stability toward reachable targets and bounded stability toward unreachable ones. Benchmark comparisons demonstrate improved accuracy when targets lie at or near singular poses.

Core claim

J-PARSE interprets the commanded end-effector velocity component-wise according to instantaneous mobility, replaces the original Jacobian with a Safety Jacobian whose manipulability ellipsoid aspect ratio stays above a chosen threshold, projects the command onto non-singular and singular subspaces, damps the singular projection by a factor derived from the threshold, and applies the right inverse of the non-singular Safety Jacobian; the resulting control law guarantees continuous signals and the stated stability properties when joint limits and collisions are absent.

What carries the argument

The Safety Jacobian, a modified Jacobian matrix constructed to keep the aspect ratio of the manipulability ellipsoid above a threshold, which supports safe projection and right-inverse computation.

If this is right

  • Safe entry into and exit from low-rank configurations without abrupt velocity changes.
  • Asymptotic stability when the target pose lies inside the reachable workspace.
  • Bounded stability when the target pose lies outside the reachable workspace.
  • Higher accuracy than prior methods when targets coincide with or pass through singular configurations.
  • Expanded usable workspace for teleoperation, visual servoing, and learning-based control.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The projection scaling may generalize to manipulators whose singularities arise from different geometric causes if the aspect-ratio threshold can be tuned accordingly.
  • Integration with joint-limit avoidance or collision-avoidance layers would be a direct next step once the core singularity handling is verified on hardware.
  • Because the method operates at the velocity level, it could be combined with existing acceleration or torque controllers without redesigning the inner loop.
  • Empirical measurement of tracking error versus threshold value on a physical arm would quantify the practical trade-off between smoothness and speed near singularities.

Load-bearing premise

A Safety Jacobian can always be formed that respects the aspect-ratio threshold yet still yields a valid continuous control signal after projection and right-inverse steps.

What would settle it

A controlled experiment or simulation in which a manipulator commanded through a known singular pose using J-PARSE produces a discontinuous joint-velocity command or diverges from the target trajectory.

Figures

Figures reproduced from arXiv: 2505.00306 by Honghao Zhen, Matthew Strong, Monroe Kennedy III, Shivani Guptasarma.

Figure 1
Figure 1. Figure 1: By appropriately modifying the commanded velocity along singular directions, the manipulator is able to move both [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Summary of the real-time J-PARSE algorithm. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A summary of the J-PARSE algorithm using the example of a planar 3-R manipulator. In this example, the end-effector (EE) twist is simply the linear velocity in the plane. When J is well-conditioned, the regular pseudoinverse is used. When the inverse condition number of J falls below γ, the safety Jacobian matrix Js (corresponding to a velocity ellipse with the same major axis as J, but aspect ratio γ) and… view at source ↗
Figure 4
Figure 4. Figure 4: Simulation results for the joint velocity q˙2 of the planar 2-R manipulator moving from a non-singular to singular configuration, controlled by the algorithms described in Table I (a) DLS and ADLS, (b) J-PARSE and EDLS, (c) J-PARSE and ADLS tuned for similar behavior. All controllers used k = 0.1, ∆t = 0.01 s. teleoperation of the XArm7 manipulator for a pick-and-place task. The same task demonstrated with… view at source ↗
Figure 5
Figure 5. Figure 5: Simulation results for the planar 2-R manipulator moving from a (a) non-singular to singular configuration identical [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Two sets of keypoints were used for the goal reaching [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Characterization experiments on a simulated X-Arm7 reaching line keypoints listed in Figure 6 with controller [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Characterization experiments on a physical X-Arm7 reaching line keypoints listed in Figure 6 with controller [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Kinova Gen3 passing through internal singularities with a human in the loop. When both q2 and q6 are close to zero, the inverse condition number drops to zero multiple times without the end-effector position being at its maximum extent. As expected with J-PARSE, joint velocities are well within reasonable limits throughout. kx = ky = 2, kz = 1; controller frequency 30 Hz. Errors are small in both position … view at source ↗
Figure 10
Figure 10. Figure 10: PUMA560 following an open-loop sinusoidal motion [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: J-PARSE compared with the default X-Arm7 controller in a pick-and-place task for an object near the boundary of the workspace. End-effector twists in SE(3) are commanded via SpaceMouse (scaling user inputs along each DoF to a range of [0, 0.08] for position and [0, 0.3] for orientation, tuned by operator preference; controller frequency 30 Hz). Before reaching the object, the default controller causes joi… view at source ↗
Figure 12
Figure 12. Figure 12: Visual servoing with the Gen3 manipulator, causing it to enter a singular configuration to reach towards a target pose outside the workspace, and then retract to a non-singular region, with joint velocities remaining within reasonable bounds throughout; kx = ky = 2, kz = 1; controller frequency 30 Hz. learning algorithms, as it shows that robust and safe policies can be learned where the robot may be requ… view at source ↗
Figure 13
Figure 13. Figure 13: Imitation Learning. A human provides 74 demonstra￾tions of commanded task space twist t; then the robot learns to condition the robot pose action, which is converted to a twist action, on the third-person camera view from that data. Above, a successful single-action demonstration is shown as proof of concept for the use of J-PARSE for imitation learning. velocity is modified – in either direction, or magn… view at source ↗
read the original abstract

J-PARSE is an algorithm for smooth first-order inverse kinematic control of a serial manipulator near kinematic singularities. The commanded end-effector velocity is interpreted component-wise, according to the available mobility in each dimension of the task space. First, a substitute "Safety" Jacobian matrix is created, keeping the aspect ratio of the manipulability ellipsoid above a threshold value. The desired motion is then projected onto non-singular and singular directions, and the latter projection scaled down by a factor informed by the threshold value. A right-inverse of the non-singular Safety Jacobian is applied to the modified command. In the absence of joint limits and collisions, this ensures safe transition into and out of low-rank configurations, guaranteeing asymptotic stability for reaching target poses within the workspace, and stability for those outside. Velocity control with J-PARSE is benchmarked against approaches from the literature, and shows high accuracy in reaching and leaving singular target poses. By expanding the available workspace of manipulators, the algorithm finds applications in teleoperation, servoing, and learning. Videos and code are available at https://jparse-manip.github.io/.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes J-PARSE, a Jacobian-based projection algorithm for smooth first-order inverse kinematic control of serial manipulators near singularities. It constructs a substitute 'Safety Jacobian' to maintain the manipulability ellipsoid aspect ratio above a user-selected threshold, projects the desired end-effector velocity onto non-singular and singular directions (scaling the latter by a factor informed by the threshold), and applies the right-inverse of the modified Jacobian. In the absence of joint limits and collisions, the method is claimed to ensure safe transitions into and out of low-rank configurations, guaranteeing asymptotic stability for target poses inside the workspace and stability for those outside. Benchmarking against literature methods reports high accuracy in reaching and leaving singular poses, with applications in teleoperation, servoing, and learning; code and videos are provided.

Significance. If the continuity of the composite map and the stability guarantees can be rigorously established, the algorithm would meaningfully expand the usable workspace of serial manipulators by enabling reliable operation near singularities. The open release of code and demonstration videos is a clear strength that supports reproducibility and practical assessment.

major comments (2)
  1. [Abstract] Abstract: the central claim that the algorithm 'guarantees asymptotic stability for reaching target poses within the workspace' is asserted without any derivation, Lyapunov analysis, error bounds, or explicit handling of the threshold choice. The post-hoc scaling informed by the threshold is described at high level only, leaving open whether the composite map (Safety Jacobian formation + projection + right-inverse) remains continuous and Lipschitz at rank-drop events.
  2. [Algorithm description (abstract)] Paragraph describing the algorithm steps: the assumption that a Safety Jacobian can always be constructed to keep the manipulability ellipsoid aspect ratio above the chosen threshold while still allowing the subsequent projection and right-inverse steps to produce a valid, continuous control signal is stated without an explicit construction rule, pseudocode, or proof that the effective mapping stays well-defined and continuous when the original Jacobian rank changes.
minor comments (2)
  1. [Abstract] The abstract introduces the 'Safety Jacobian' and 'aspect ratio threshold' without a forward reference to the section where their explicit construction is defined; adding such a pointer would improve readability.
  2. [Abstract] Benchmarking results are summarized as 'high accuracy' but no quantitative metrics (e.g., mean error, success rate, or comparison tables) are mentioned in the abstract; including a brief numerical summary would strengthen the claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review. The comments highlight important points regarding the rigor of our stability claims and the explicitness of the algorithm's continuity properties. We address each major comment below and will revise the manuscript to incorporate additional analysis, pseudocode, and proofs as outlined.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the algorithm 'guarantees asymptotic stability for reaching target poses within the workspace' is asserted without any derivation, Lyapunov analysis, error bounds, or explicit handling of the threshold choice. The post-hoc scaling informed by the threshold is described at high level only, leaving open whether the composite map (Safety Jacobian formation + projection + right-inverse) remains continuous and Lipschitz at rank-drop events.

    Authors: We agree that the abstract presents the stability guarantee concisely without a full derivation. The manuscript body (Section 4) sketches the argument that the Safety Jacobian remains full-rank by construction and that the directional projection prevents unbounded joint velocities, leading to asymptotic convergence inside the workspace. However, we acknowledge the absence of an explicit Lyapunov analysis and bounds on the threshold. In the revision we will add a dedicated stability subsection deriving a Lyapunov function V = (1/2) ||e||^2 (with e the task-space error) and showing that the modified velocity command yields dot{V} <= -k ||e||^2 for a positive k dependent on the threshold. We will also prove local Lipschitz continuity of the composite map by bounding the variation of the SVD-based projection across rank transitions. revision: yes

  2. Referee: [Algorithm description (abstract)] Paragraph describing the algorithm steps: the assumption that a Safety Jacobian can always be constructed to keep the manipulability ellipsoid aspect ratio above the chosen threshold while still allowing the subsequent projection and right-inverse steps to produce a valid, continuous control signal is stated without an explicit construction rule, pseudocode, or proof that the effective mapping stays well-defined and continuous when the original Jacobian rank changes.

    Authors: Section 3 of the manuscript describes the Safety Jacobian via singular-value thresholding to enforce the aspect-ratio bound, followed by projection onto the resulting singular directions. We accept that an explicit algorithmic rule and continuity proof are not provided at the level of detail requested. The revision will include (i) pseudocode for the full J-PARSE procedure, (ii) a formal argument that the thresholding operator is continuous in the Jacobian entries, and (iii) a proof in the appendix that the overall map remains well-defined and continuous at rank-drop events by showing that the scaled singular values and the associated projectors converge to the limiting case. revision: yes

Circularity Check

0 steps flagged

No circularity: algorithm definition and stability claim remain independent of inputs

full rationale

The paper presents J-PARSE as a sequence of explicit matrix operations (Safety Jacobian construction to enforce aspect-ratio threshold, directional projection, scaling of singular components, and right-inverse application). The asymptotic-stability guarantee is asserted as a consequence of these operations under the stated assumptions (no joint limits or collisions), without any equation that reduces the stability result to a fitted parameter, a self-referential definition, or a load-bearing self-citation. No step matches the enumerated circularity patterns; the threshold is an explicit user parameter rather than a fitted quantity renamed as a prediction. The derivation therefore contains independent content and is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central claim rests on a user-chosen threshold parameter that defines the Safety Jacobian and on the domain assumption of no joint limits or collisions for the stability guarantee.

free parameters (1)
  • aspect ratio threshold
    Determines when the Safety Jacobian is substituted and how much to scale the singular velocity projection; chosen to maintain ellipsoid shape but not derived from first principles.
axioms (1)
  • domain assumption The robot is a serial manipulator operating without joint limits or collisions.
    Explicitly invoked in the abstract when stating the stability guarantees.
invented entities (1)
  • Safety Jacobian no independent evidence
    purpose: Substitute matrix that preserves a minimum manipulability ellipsoid aspect ratio to avoid singularity issues.
    Newly introduced construct whose properties are used to modify the velocity command.

pith-pipeline@v0.9.0 · 5740 in / 1315 out tokens · 54119 ms · 2026-05-22T18:21:47.405508+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

47 extracted references · 47 canonical work pages · 4 internal anchors

  1. [1]

    Kinova Robotics,Kinova® Jaco Assistive Robot User Guide, 2021, Accessed: 2025-10-14, p. 18. [Online]. Available: https://assistive.kinovarobotics.com/ uploads/EN-UG-007-Jaco-user-guide-R05.pdf

    work page 2021

  2. [2]

    UFACTORY,xArm User Manual V2.0.0, 2023, Accessed: 2025-03-22, pp. 189–191. [Online]. Available: https://www.ufactory.cc/wp-content/uploads/ 2023/05/xArm-User-Manual-V2.0.0.pdf

    work page 2023

  3. [3]

    I. A. Sucan and S. Chitta, “Moveit.” [Online]. Available: moveit.ros.org

    work page

  4. [4]

    Moveit servo,

    B. O’Neil, A. Zelenak, and B. An- derson, “Moveit servo,” 2019. [Online]. Available: https://github.com/moveit/moveit/blob/master/ moveit ros/moveit servo/src/servo calcs.cpp#L694

    work page 2019

  5. [5]

    Moveit 2 moveit servo,

    A. Zelenak and V . M. Ibrahim, “Moveit 2 moveit servo,” 2023. [Online]. Available: https://github.com/moveit/moveit2/blob/main/ moveit ros/moveit servo/src/utils/common.cpp#L282

    work page 2023

  6. [6]

    Real-time obstacle avoidance for manipu- lators and mobile robots,

    O. Khatib, “Real-time obstacle avoidance for manipu- lators and mobile robots,” in1985 IEEE International Conference on Robotics and Automation Proceedings, vol. 2, Mar. 1985, pp. 500–505

    work page 1985

  7. [7]

    Augmented haptics of manipulator kinematic condition,

    T. Maneewarn and B. Hannaford, “Augmented haptics of manipulator kinematic condition,” inTelemanipulator and Telepresence Technologies VI, vol. 3840. SPIE, 1999, pp. 54–64

    work page 1999

  8. [8]

    A framework for singularity-robust manipulator control during physical human-robot interaction,

    M. G. Carmichael, D. Liu, and K. J. Waldron, “A framework for singularity-robust manipulator control during physical human-robot interaction,”The International Journal of Robotics Research, vol. 36, no. 5-7, pp. 861–876, 2017. [Online]. Available: https://doi.org/10.1177/0278364917698748

    work page doi:10.1177/0278364917698748 2017

  9. [9]

    Fast and ro- bust inverse kinematics of serial robots using halley’s method,

    S. Lloyd, R. A. Irani, and M. Ahmadi, “Fast and ro- bust inverse kinematics of serial robots using halley’s method,”IEEE Transactions on Robotics, vol. 38, no. 5, pp. 2768–2780, 2022

    work page 2022

  10. [10]

    Manipulator control at kine- matic singularities: A dynamically consistent strategy,

    K.-S. Chang and O. Khatib, “Manipulator control at kine- matic singularities: A dynamically consistent strategy,” in Proceedings 1995 IEEE/RSJ International Conference on Intelligent Robots and Systems. Human Robot Interaction and Cooperative Robots, vol. 3, Aug. 1995, pp. 84–88 vol.3

    work page 1995

  11. [11]

    Learning operational space control,

    J. Peters and S. Schaal, “Learning operational space control,” inRobotics Science and Systems 2006, 08 2006

    work page 2006

  12. [12]

    Conti- nuity of varying-feature-set control laws,

    N. Mansard, A. Remazeilles, and F. Chaumette, “Conti- nuity of varying-feature-set control laws,”IEEE Trans- actions on Automatic Control, vol. 54, no. 11, pp. 2493– 2505, 2009

    work page 2009

  13. [13]

    A unified approach to integrate unilateral constraints in the stack of tasks,

    N. Mansard, O. Khatib, and A. Kheddar, “A unified approach to integrate unilateral constraints in the stack of tasks,”IEEE Transactions on Robotics, vol. 25, no. 3, pp. 670–685, 2009

    work page 2009

  14. [14]

    A continuous task transition algorithm for operational space control framework,

    H. Han, J. Lee, and J. Park, “A continuous task transition algorithm for operational space control framework,” in 2012 9th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI), 2012, pp. 148–152

    work page 2012

  15. [15]

    Robot control near singularity and joint limit using a continuous task transition algorithm,

    H. Han and J. Park, “Robot control near singularity and joint limit using a continuous task transition algorithm,” International Journal of Advanced Robotic Systems, vol. 10, no. 10, p. 346, 2013. [Online]. Available: https://doi.org/10.5772/56714

    work page doi:10.5772/56714 2013

  16. [16]

    Inverse Kinematic Solu- tions With Singularity Robustness for Robot Manipulator Control,

    Y . Nakamura and H. Hanafusa, “Inverse Kinematic Solu- tions With Singularity Robustness for Robot Manipulator Control,”Journal of Dynamic Systems, Measurement, and Control, vol. 108, no. 3, pp. 163–171, Sept. 1986

    work page 1986

  17. [17]

    Manipulator Inverse Kinematic Solu- 20 tions Based on Vector Formulations and Damped Least- Squares Methods,

    C. W. Wampler, “Manipulator Inverse Kinematic Solu- 20 tions Based on Vector Formulations and Damped Least- Squares Methods,”IEEE Transactions on Systems, Man, and Cybernetics, vol. 16, no. 1, pp. 93–101, Jan. 1986

    work page 1986

  18. [18]

    Overview of damped least-squares methods for inverse kinematics of robot manipulators,

    A. S. Deo and I. D. Walker, “Overview of damped least-squares methods for inverse kinematics of robot manipulators,”Journal of Intelligent and Robotic Systems, vol. 14, pp. 43–68, 1995. [Online]. Available: https://doi.org/10.1007/BF01254007

    work page doi:10.1007/bf01254007 1995

  19. [19]

    Human preferences in using damping to manage sin- gularities during physical human-robot collaboration,

    M. G. Carmichael, R. Khonasty, S. Aldini, and D. Liu, “Human preferences in using damping to manage sin- gularities during physical human-robot collaboration,” in 2020 IEEE International Conference on Robotics and Automation (ICRA), 2020, pp. 10 184–10 190

    work page 2020

  20. [20]

    Automatic generation of kine- matics for a reconfigurable modular manipulator system,

    L. Kelmar and P. Khosla, “Automatic generation of kine- matics for a reconfigurable modular manipulator system,” inProceedings. 1988 IEEE International Conference on Robotics and Automation, 1988, pp. 663–668 vol.2

    work page 1988

  21. [21]

    General inverse kinematics with the error damped pseudoinverse,

    S. Chan and P. Lawrence, “General inverse kinematics with the error damped pseudoinverse,” inProceedings. 1988 IEEE International Conference on Robotics and Automation, 1988, pp. 834–839 vol.2

    work page 1988

  22. [22]

    Selectively damped least squares for inverse kinematics,

    S. R. Buss and J.-S. Kim, “Selectively damped least squares for inverse kinematics,”Journal of Graphics Tools, vol. 10, pp. 37 – 49, 2005. [Online]. Available: https://api.semanticscholar.org/CorpusID:6860165

    work page 2005

  23. [23]

    D. Vito, C. Natale, and G. Antonelli, “A comparison of damped least squares algorithms for inverse kinematics of robot manipulators * *this work was supported by the european community through theprojectsrobust(h2020- 690416),euroc(fp7-608849), dexrov (h2020-635491) and aeroarms (h2020-644271).”IFAC-PapersOnLine, vol. 50, pp. 6869–6874, 07 2017

    work page 2017

  24. [24]

    Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware

    T. Z. Zhao, V . Kumar, S. Levine, and C. Finn, “Learning fine-grained bimanual manipulation with low-cost hard- ware,”arXiv preprint arXiv:2304.13705, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  25. [25]

    $\pi_0$: A Vision-Language-Action Flow Model for General Robot Control

    K. Black, N. Brown, D. Driess, A. Esmail, M. Equi, C. Finn, N. Fusai, L. Groom, K. Hausman, B. Ichter, et al., “π 0: A vision-language-action flow model for general robot control,”arXiv preprint arXiv:2410.24164, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  26. [26]

    OpenVLA: An Open-Source Vision-Language-Action Model

    M. J. Kim, K. Pertsch, S. Karamcheti, T. Xiao, A. Balakr- ishna, S. Nair, R. Rafailov, E. Foster, G. Lam, P. Sanketi, et al., “Openvla: An open-source vision-language-action model,”arXiv preprint arXiv:2406.09246, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  27. [27]

    Diffusion pol- icy: Visuomotor policy learning via action diffusion,

    C. Chi, Z. Xu, S. Feng, E. Cousineau, Y . Du, B. Burchfiel, R. Tedrake, and S. Song, “Diffusion pol- icy: Visuomotor policy learning via action diffusion,” The International Journal of Robotics Research, p. 02783649241273668, 2023

    work page 2023

  28. [28]

    What Matters in Learning from Offline Human Demonstrations for Robot Manipulation

    A. Mandlekar, D. Xu, J. Wong, S. Nasiriany, C. Wang, R. Kulkarni, L. Fei-Fei, S. Savarese, Y . Zhu, and R. Mart´ın-Mart´ın, “What matters in learning from offline human demonstrations for robot manipulation,”arXiv preprint arXiv:2108.03298, 2021

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  29. [29]

    Keypoint action tokens enable in-context imitation learning in robotics,

    N. Di Palo and E. Johns, “Keypoint action tokens enable in-context imitation learning in robotics,”arXiv preprint arXiv:2403.19578, 2024

    work page arXiv 2024

  30. [30]

    Open x-embodiment: Robotic learning datasets and rt-x models: Open x-embodiment collaboration 0,

    A. O’Neill, A. Rehman, A. Maddukuri, A. Gupta, A. Padalkar, A. Lee, A. Pooley, A. Gupta, A. Man- dlekar, A. Jain,et al., “Open x-embodiment: Robotic learning datasets and rt-x models: Open x-embodiment collaboration 0,” in2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 6892–6903

    work page 2024

  31. [31]

    Storm: An in- tegrated framework for fast joint-space model-predictive control for reactive manipulation,

    M. Bhardwaj, B. Sundaralingam, A. Mousavian, N. D. Ratliff, D. Fox, F. Ramos, and B. Boots, “Storm: An in- tegrated framework for fast joint-space model-predictive control for reactive manipulation,” inConference on Robot Learning. PMLR, 2022, pp. 750–759

    work page 2022

  32. [32]

    Learning singularity avoidance,

    J. Manavalan and M. Howard, “Learning singularity avoidance,” in2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2019, pp. 6849–6854

    work page 2019

  33. [33]

    Bunny-visionpro: Real-time bimanual dexterous teleoperation for imitation learning,

    R. Ding, Y . Qin, J. Zhu, C. Jia, S. Yang, R. Yang, X. Qi, and X. Wang, “Bunny-visionpro: Real-time bimanual dexterous teleoperation for imitation learning,”arXiv preprint arXiv:2407.03162, 2024

    work page arXiv 2024

  34. [34]

    Kinova Robotics,Kinova® Gen3 Ultra Lightweight Robot User Guide, 2022, Accessed: 2025-03-22, p. 95. [Online]. Available: https://www.kinovarobotics. com/uploads/User-Guide-Gen3-R07.pdf

    work page 2022

  35. [35]

    A review of cuspidal serial and parallel manipulators,

    P. Wenger and D. Chablat, “A review of cuspidal serial and parallel manipulators,”Journal of Mechanisms and Robotics, vol. 15, pp. 1–37, 09 2022

    work page 2022

  36. [36]

    Liberzon,Switching in systems and control

    D. Liberzon,Switching in systems and control. Springer, 2003, vol. 190

    work page 2003

  37. [37]

    Estimating the Jacobian of the Singular Value Decomposition : Theory and Applications,

    T. Papadopoulo and M. L. Lourakis, “Estimating the Jacobian of the Singular Value Decomposition : Theory and Applications,” INRIA, Tech. Rep. R-3961, inria- 00072686, 2000

    work page 2000

  38. [38]

    Notes on partial differential equations, Department of Mathematics, UC Davis,

    J. K. Hunter, “Notes on partial differential equations, Department of Mathematics, UC Davis,” 2014. [Online]. Available: https://www.math.ucdavis.edu/ ∼hunter/pdes/ pde notes.pdf

    work page 2014

  39. [39]

    Inverse kinematic algorithms for redundant systems,

    H. Das, J.-E. Slotine, and T. Sheridan, “Inverse kinematic algorithms for redundant systems,” inProceedings. 1988 IEEE International Conference on Robotics and Automa- tion, 1988, pp. 43–48 vol.1

    work page 1988

  40. [40]

    Design and use paradigms for gazebo, an open-source multi-robot simulator,

    N. Koenig and A. Howard, “Design and use paradigms for gazebo, an open-source multi-robot simulator,” in Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Sendai, Japan, September 2004, pp. 2149–2154

    work page 2004

  41. [41]

    The Mathematics of Coordinated Con- trol of Prosthetic Arms and Manipulators,

    D. E. Whitney, “The Mathematics of Coordinated Con- trol of Prosthetic Arms and Manipulators,”Journal of Dynamic Systems, Measurement, and Control, vol. 94, no. 4, pp. 303–309, Dec. 1972

    work page 1972

  42. [42]

    Kinemat- ically Redundant Manipulators,

    S. Chiaverini, G. Oriolo, and I. D. Walker, “Kinemat- ically Redundant Manipulators,” inSpringer Handbook of Robotics, B. Siciliano and O. Khatib, Eds. Berlin, Heidelberg: Springer, 2008, pp. 245–268

    work page 2008

  43. [43]

    Apriltag: A robust and flexible visual fiducial system,

    E. Olson, “Apriltag: A robust and flexible visual fiducial system,” in2011 IEEE International Conference on Robotics and Automation, 2011, pp. 3400–3407

    work page 2011

  44. [44]

    Denoising Diffusion Probabilistic Models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising Diffusion Probabilistic Models,” inAdvances in Neural Information 21 Processing Systems, vol. 33. Curran Associates, Inc., 2020, pp. 6840–6851

    work page 2020

  45. [45]

    Diffusion pol- icy: Visuomotor policy learning via action diffusion,

    C. Chi, Z. Xu, S. Feng, E. Cousineau, Y . Du, B. Burchfiel, R. Tedrake, and S. Song, “Diffusion pol- icy: Visuomotor policy learning via action diffusion,” The International Journal of Robotics Research, p. 02783649241273668, Oct. 2024

    work page 2024

  46. [46]

    FiLM: Visual Reasoning with a General Conditioning Layer,

    E. Perez, F. Strub, H. de Vries, V . Dumoulin, and A. Courville, “FiLM: Visual Reasoning with a General Conditioning Layer,”Proceedings of the AAAI Confer- ence on Artificial Intelligence, vol. 32, no. 1, Apr. 2018

    work page 2018

  47. [47]

    Johnston,Advanced Linear and Matrix Algebra

    N. Johnston,Advanced Linear and Matrix Algebra. Springer International Publishing, 2021, pp. 200–201. [Online]. Available: https://www.springer.com/us/book/ 9783030528140

    work page 2021