WorksComputer Science
Geometry-Aware Humanoid Retargeting via Manifold ODEs
Abstract
In this paper, we focus on interaction-preserving motion retargeting, which transfers human demonstrations to humanoids while maintaining spatial relationships with objects and terrain. Previous work, such as OmniRetarget, addresses this task through temporally coupled constrained optimization, requiring repeated sequential quadratic programming (SQP) iterations at each frame. We study this problem from an ordinary differential equation (ODE) perspective by modeling the continuous-time limit of its framewise Karush-Kuhn-Tucker (KKT) conditions as an ODE on the robot configuration manifold. This viewpoint offers two benefits: (i) it connects OmniRetarget's SQP iterations to a backward discretization of the ODE, enabling us to establish their local convergence under suitable regularity assumptions; and (ii) it leads to TaSP-KKT (tangent-space parametrization for KKT flow), a new retargeting method based on a forward discretization of the KKT flow. The proposed method achieves the same first-order convergence to the ODE solution as the corresponding backward scheme while requiring only one weighted tangent-cone projection and a closed-form manifold update per step, thereby avoiding repeated inner SQP iterations. Numerical experiments show that TaSP-KKT achieves retargeting quality comparable to, though slightly lower than, that of OmniRetarget while running approximately twice as fast.
Introduction
Whole-body tasks such as carrying objects and stepping onto a platform require humanoid robots to coordinate their limbs while maintaining contact with objects and the environment [1, 2]. Human demonstrations provide examples of this coordination and can serve as references for learning robot behaviors. However, differences in body proportions, joint structure, and motion limits prevent human motions from being transferred directly to a robot. Generating usable robot motion references therefore requires adapting both the demonstrated movements and their interactions with the scene to the robot’s embodiment. Motion retargeting aims to enable a target character or robot to reproduce a source motion despite differences in body proportions or skeletal structure while preserving the motion’s essential characteristics.
Early approaches to motion retargeting include the space-time optimization of Gleicher [3] and the inverse-rate formulation of Choi and Ko [4]. More recently, General Motion Retargeting (GMR) [5] combines local scaling with differential inverse kinematics to generate humanoid motion references.
Interaction-preserving motion retargeting, which maintains spatial relationships and contacts during motion transfer, has also received much attention. For example, Contact-Aware Retargeting [6] combines a geometry-conditioned recurrent network with encoder-space optimization to reduce interpenetration and preserve self-contacts and ground contacts. RET [7] uses a learned model to predict residual motion corrections that account for skeletal relationships and body geometry, aiming to reduce interpenetration and missing contacts when transferring motion between characters with different skeletal structures.
Among the most recent work is OmniRetarget [8], which uses an interaction mesh to represent spatial relationships among the body, objects, and terrain. At each frame, it minimizes Laplacian mesh deformation subject to kinematic constraints using a sequential quadratic programming (SQP) solver. The resulting references reduced penetration and foot skating and supported object carrying and terrain traversal on a Unitree G1 humanoid.
In this paper, we study humanoid motion retargeting and OmniRetarget from the perspective of ordinary differential equations (ODEs). Our motivation is twofold. First, OmniRetarget solves a sequence of temporally coupled constrained optimization problems, but the original work does not establish convergence guarantees for its SQP iterations. The longstanding connections between optimization and ODEs provide tools for analyzing convergence [9, 10, 11]. These connections motivate us to establish theoretical guarantees for OmniRetarget through a dynamical formulation.
Second, an ODE formulation can guide the development of new algorithms. In deep learning, Haber et al. [12] use a continuous-time interpretation of residual networks to develop multiscale training methods, which attains higher classification accuracy and lower training cost. In diffusion generative modeling, Lu et al. [13] exploit the structure of diffusion ODEs to derive DPM-Solver, which use with high-order ODE schemes to accelerate sampling. These examples motivate using the structure of retargeting dynamics to design more efficient numerical methods.
ODE-based methods have also been explored in humanoid planning and retargeting. Klein et al. [14] generate robot arm motions as piecewise geodesics governed by second-order ODEs under an inertia-based Riemannian metric. Their subsequent work [15] modifies this metric using barrier functions to generate collision-avoiding trajectories. Li et al. [16] use a learned conditional ODE flow to adapt motions across characters with the same skeletal topology but different body proportions. MoReFlow [17] formulates human-to-robot motion mapping as an ODE flow in latent space and uses flow matching to train a neural network that parameterizes its velocity field.
In this paper, we model the evolution of the optimal robot configurations generated by OmniRetarget as an ODE on , where represents the floating-base orientation using a unit quaternion, and contains the base translation and joint coordinates. We then use this ODE formulation to analyze the convergence of OmniRetarget’s SQP solver and derive a computationally efficient retargeting algorithm. Our contributions are summarized as follows:
We develop a manifold ODE formulation of OmniRetarget. In particular, we characterize the evolution of OmniRetarget’s framewise KKT point as a projected flow on the robot configuration manifold . The formulation provides a common basis for analyzing the existing solver and designing new retargeting algorithms.
We connect OmniRetarget’s SQP iterations to a backward discretization of the KKT flow. Through this connection, we interpret the SQP iterations as Newton steps for the backward discretization and establish local convergence of the SQP solver.
We propose TaSP-KKT, a retargeting algorithm based on a forward discretization of the KKT flow. It achieves the same convergence order as the SQP-based backward scheme. Each step requires only one weighted tangent-cone projection and a closed-form manifold update, avoiding repeated inner SQP iterations and thus reducing the computational cost. Preliminary numerical experiments show that TaSP-KKT achieves retargeting quality comparable to that of OmniRetarget while running approximately twice as fast.
Preliminaries
Ordinary Differential Equations
An ODE describes the evolution of a state over continuous time . At each time , the vector field assigns a rate of change to the current state. Together with an initial state , the ODE defines the initial-value problem
where . A solution of (1) on is a trajectory that starts at and whose derivative agrees with the vector field at every time.
a) Discrete approximation: To approximate the continuous-time trajectory numerically, we discretize time and advance the state in finite steps. For a step size and sampling times , a one-step method generates approximations according to
where denotes the numerical update map. For example, the forward and backward Euler methods are given by
Forward Euler evaluates the vector field at the known current state and thus provides an explicit update. Backward Euler evaluates the vector field at the unknown next state and therefore requires solving an implicit equation, which is generally nonlinear [18].
b) Convergence order: Convergence order quantifies how rapidly the numerical error decreases as the step size tends to zero over a fixed time interval . Let . A method has global convergence order if, for all sufficiently small ,
where the constant may depend on but is independent of . Under standard smoothness assumptions, both Euler methods have first-order convergence [18].
ODEs on Robot Configuration Space
We represent the configuration of a floating-base robot as
The unit quaternion represents the base orientation, while contains the base translation and joint coordinates. Quaternion signs are chosen consistently across frames. The tangent space consists of the velocities of smooth curves on passing through . Differentiating the constraint shows that a quaternion velocity must satisfy . Thus, a tangent quaternion vector is perpendicular to , and the tangent space of at is given by
An ODE on the configuration manifold must generate tangent velocities [19 Ch. IV]:
Equivalently,
Consider two unit quaternions in the same hemisphere. Let denote the orthogonal projection of their displacement onto the tangent space at :
The point can therefore be decomposed as , where . Since and have unit norm,
Selecting the positive root gives . Thus, for any with , the tangent projection uniquely determines within the hemisphere through the tangent space parametrization (TaSP) map [20]:
Combining this parametrization with forward Euler yields the TaSP–Euler method for (6). Starting from , we set and compute using
The update preserves when . The forward TaSP–Euler method has first-order convergence [18], i.e.,
A gradient flow for a smooth function evolves along the negative Riemannian gradient of :
where denotes the Riemannian gradient, which is obtained by projecting the Euclidean gradient onto . For , the orthogonal projector is
which gives
Here, denotes the Euclidean gradient, and is its Riemannian gradient with respect to the quaternion component.
SQP-based Motion Retargeting
OmniRetarget [8] transfers human motion to a robot by preserving the spatial relationships among the body, manipulated objects, and terrain. Consider a human demonstration over the time interval , uniformly sampled with a sampling period . The sampled demonstration is represented by a sequence , where contains the 3D positions of human body keypoints and points sampled from object and terrain surfaces. At each sample , OmniRetarget constructs an interaction mesh by applying Delaunay tetrahedralization [21] to . Let denote the associated graph Laplacian [22], defined such that the th row of represents the displacement of vertex from the mean position of its neighbors [8].
Let denote the retargeted robot configuration at frame . The map returns the corresponding target mesh vertices by combining robot keypoints computed through forward kinematics with prescribed object and terrain points. At each frame, OmniRetarget obtains the desired robot configuration by solving the following constrained optimization problem
The objective and constraints have the following roles:
Interaction preservation. The loss is defined as
where stacks the columns of a matrix. The residual compares corresponding vertices’ offsets from their neighbors’ average positions in the robot and human interaction meshes. Minimizing therefore encourages preservation of local spatial relationships among the body, objects, and terrain.
Temporal regularization. With , define , where is symmetric positive semidefinite. This penalty discourages large changes between consecutive frames, encouraging temporal smoothness [8].
Kinematic feasibility. At time step , the vector collects the inequality constraints enforcing non-penetration and joint position and velocity limits. Similarly, collects the equality constraints, including those specifying stance-foot positions. The subscript indicates the time dependence of these constraint functions.
Note that our temporal penalty includes an explicit factor , which is absent from the original formulation [8]. The two expressions differ only in their normalization and are equivalent under an appropriate rescaling of the weight matrix . OmniRetarget solves (13) using SQP. At frame , the iterations are initialized with the SQP solution from frame , i.e., . At iteration , the SQP process linearizes the interaction residual and constraint functions at to obtain the following trust-region subproblem for the increment
Here, denotes the Jacobian of the residual , while and denote the row Jacobians of the scalar constraints and , respectively. All Jacobians are evaluated with respect to at , with the frame data held fixed.
After solving (15), we decompose the resulting increment into its quaternion and Euclidean components, . The next iterate is obtained by adding the increment to the current iterate and normalizing the quaternion component:
After SQP iterations, the output for frame is defined as .
KKT Flow of Humanoid Motion Retargeting
In this section, we model the evolution of the framewise KKT points of (13) as a constrained gradient ODE on in the limit as the sampling period . As a result, starting from a KKT point at the initial frame, one can integrate the proposed dynamics to track the corresponding KKT trajectory, thereby avoiding the need to solve a separate optimization problem at each subsequent frame. Set and write , , and . Define the feasible set and the active inequality index set as
At , write and define the active constraint Jacobian
Write and for the outward normal and tangent cones to relative to . At a feasible point , these cones admit the representations [23]
Since is a nonempty closed convex cone, any matrix defines a unique weighted projection
Given any at which has full row rank, any , any , the projection is characterized by
For analysis, we make the following regularity and the linear independence constraint qualification (LICQ) assumptions [18, 23]. In Assumptions 1–3, denotes any of the scalar functions , (), or ().
Assumption 1 (Boundedness): There exists a constant , independent of , such that
holds for every frame and every function .
Assumption 2 (Lipschitz continuity): There exists a constant , independent of , such that
holds for every frame and every function .
Assumption 3 (L-smoothness): There exists a constant , independent of , such that
holds for every frame and every function (note that here the Riemannian gradients are compared as vectors in the ambient space ).
Assumption 4 (LICQ): For every frame and every KKT point , the active constraint Jacobian has full row rank.
Under these assumptions, we model the evolution of the framewise KKT solutions as a constrained gradient ODE, as formalized in the following theorem.
Theorem 1 (KKT flows): Suppose Assumptions 1–4 hold, and . Let . For each , let be a KKT point of (13) at , with feasible configuration . As , these KKT conditions yield
The proof is in Appendix I. Theorem 1 suggests computing the entire retargeted motion by solving a single initial-value problem. Starting from , integrating the projected flow over the full motion interval yields a configuration trajectory whose sampled states provide the retargeted frames. Thus, humanoid motion retargeting can be performed by integrating one ODE instead of solving a separate optimization problem at every frame.
Discrete Approximation of the KKT Flow
To compute the KKT trajectory numerically, we construct discrete approximations of the KKT flow. Define the projected velocity field
For a step size and a time grid , the backward and forward TaSP–Euler updates for (23) are, respectively,
and
In this section, we first establish first-order convergence of the backward TaSP–Euler method (25). We then characterize the backward update by a nonlinear system on . For a fixed reference configuration , we show that the SQP iterations (15) converge locally and that their limit differs from a nearby feasible solution of the backward system by . Building on the fact that the forward TaSP–Euler method has already been proved to converge with order one [24, 25], we propose TaSP-KKT, an algorithm for OmniRetargeting based on the update in (26). By evaluating the velocity field at the current state and advancing the configuration explicitly, TaSP-KKT avoids the nonlinear solve and associated Newton iterations required by the backward formulation, thereby reducing the computational cost per step.
Backward Euler Method and SQP Procedure
We establish the convergence order of the backward TaSP–Euler method.
Lemma 1. Suppose that Assumptions 1–4 hold, and let be an exact solution of (23). Then, for all sufficiently small step sizes , the backward TaSP–Euler iterates initialized at satisfy the first-order global error bound
where .
The proof is provided in Appendix II. We next express the implicit backward TaSP update in (25) as an equivalent nonlinear system.
Lemma 2. Suppose that Assumptions 1–4 hold. Let , , and . Set , and set . If , then satisfies the backward TaSP update (25) if and only if there exists a multiplier vector such that
The proof is in Appendix III.
Theorem 2 (SQP Convergence). Fix a reference configuration and the frame data at . Let , , and , with the index set held fixed. Set in (15), solve each SQP subproblem exactly, and apply the full normalization update (16). For sufficiently small , the SQP iterates converge to a limit
moreover, there exists a constant , independent of , such that
The proof is left in Appendix IV. Theorem 2 shows that, for a fixed reference configuration , the SQP iterations converge locally to a limit within of . By Lemma 1, the backward iterates approximate the KKT flow (23) with global error. This connection provides a KKT-flow interpretation of the SQP limit together with a local convergence guarantee for the SQP iterations.
Forward TaSP–Euler: The TaSP-KKT Method
We first state the convergence order of the forward TaSP–Euler update (26).
Theorem 3 (First-order convergence of forward TaSP). Suppose that Assumptions 1–4 hold. Let be an exact solution of (23). Then the forward TaSP–Euler iterates in (26), initialized at , satisfy
for all sufficiently small $\eta>0.
Proof. See, for example, [24], [25]. □
| Algorithm 1 TaSP-KKT: Forward Integration of the KKT Flow |
| Require: Initial state , , step size , horizon $T |
| Ensure: Configurations |
| 1: |
| 2: for do |
| 3: |
| 4: |
| 5: Decompose |
| 6: |
| 7: |
| 8: |
| 9: |
| 10: end for |
| 11: return |
Table 1.
Applying the forward TaSP–Euler method to the KKT flow yields an explicit algorithm, which we call TaSP-KKT (tangent-space parametrization for KKT flow). We summarize TaSP-KKT in Algorithm 1.
TaSP-KKT uses only the current frame data and configuration. Each integration step requires one weighted tangent-cone projection and a closed-form TaSP update, avoiding the repeated linearizations and QP solves within the framewise SQP loop. By Lemma 1 and Theorem 3, forward and backward TaSP–Euler both have first-order global accuracy under their stated assumptions. Thus, TaSP-KKT reduces the computational work per integration step while retaining the same convergence order.
Experiments and Discussions
In this section, we validate our theoretical findings via some numerical experiments.
Experimental Setup
We evaluate TaSP-KKT for retargeting human–object interaction motions to the Unitree G1 humanoid. Human demonstrations are drawn from sequences involving the large box object (largebox) in the OMOMO dataset [26], with a sampling interval of s. Using the dataset’s action annotations,1 we group the selected motions into seven patterns: Lift–Move–Place, Lift–Rotate–Replace, Push–Replace, Kick–Replace, Kick–Lift–Move–Place, Push–Release–Pull–Replace, and Pull–Replace. For each pattern, we select three complete sequences from three different subjects, yielding 21 sequences in total.
We evaluate both interaction preservation and computational efficiency. The first metric is the mean interaction loss, computed by averaging the framewise loss within each sequence and then taking an equally weighted average across the three sequences in each motion pattern:
where is the number of frames in sequence , is the retargeted robot configuration at frame , and is the corresponding interaction loss. A lower indicates better preservation of the reference interaction structure.
The second metric is the total retargeting solve time, , where denotes the solve time for the complete sequence . Since sequence lengths may differ across motion patterns, we compare solve times between methods using the same sequences within each pattern.
Results and Discussion
As shown in Figure 1, TaSP-KKT requires approximately 38%–63% less computation time than OmniRetarget while yielding slightly higher mean interaction loss. The absolute differences in mean interaction loss are on the order of for most motion patterns. These results illustrate our theoretical findings.

Figure 1. OMOMO largebox results: (a) mean interaction loss; (b) total solve time.
Conclusion
In this paper, we have presented an ODE perspective on interaction-preserving motion retargeting by characterizing the continuous-time limit of OmniRetarget’s framewise KKT conditions on the robot configuration manifold. This perspective supports both convergence analysis and algorithm design. By connecting the backward TaSP–Euler discretization to OmniRetarget’s SQP iterations, we establish a local convergence guarantee for the solver under suitable regularity assumptions. By discretizing the same KKT flow with a forward TaSP–Euler scheme, we derive TaSP-KKT, which achieves the same first-order convergence to the ODE solution as the backward method under the stated assumptions. Its explicit manifold update preserves the unit quaternion constraint for admissible step sizes, while replacing repeated inner SQP iterations with one weighted tangent-cone projection and a closed-form update per step. Numerical experiments demonstrate that TaSP-KKT runs approximately twice as fast as OmniRetarget while maintaining comparable retargeting quality, with a slight increase in mean interaction loss on the order of for most motion patterns.
In future work, we will evaluate TaSP-KKT on a broader range of datasets and assess its effectiveness in downstream tasks, such as humanoid control based on reinforcement learning.
Appendix I
Proof of Theorem 1
The manifold KKT conditions [23] retain the usual primal feasibility, dual feasibility, and complementary slackness conditions. Stationarity is expressed in the tangent space using Riemannian gradients. For the fixed reference configuration , the temporal penalty has Euclidean and Riemannian gradients
Evaluating these expressions at gives the framewise KKT conditions
Here, stacks the active inequality multipliers and the equality multipliers. Taking and , we have
Therefore, we have
Set . Since the trajectory is feasible, both the forward and backward tangent directions, and , must belong to . The polar-cone characterization therefore gives
For every , both and are tangent to . Hence (33) implies
This is the first-order optimality condition for minimizing over the tangent cone. Since , the minimizer is unique. Thus, (33) yields
The proof is completed.
Proof of Lemma 1
Proof. For brevity, write and . The boundedness and Lipschitz continuity of , , and imply that there exist constants and , independent of and , such that
The backward TaSP update (25) gives
Consequently,
where
For , rationalizing the square root yields
We introduce the frozen-field implicit sequence
This iteration has the form of the classical backward Euler method, with the velocity evaluated at the backward TaSP iterate . Subtracting the two recursions and summing gives . Hence, for ,
Next, we interpolate the auxiliary sequence linearly, i.e.,
The velocity bound implies . Therefore,
Let . Since and ,
It follows that . Since , the mean value estimate gives
uniformly in and . Thus, for ,
where is independent of , , and .
Because , the integral representations of and give
Gronwall’s inequality therefore yields
At the grid points,
Thus (27) holds with
which is independent of . □
Proof of Lemma 2
Proof. Set , . TaSP inverts tangent projection, so (25)
Applying (21) with and , there exist such that
Together with , these relations complete the proof. □
Proof of Theorem 2
Proof. Fix the reference configuration . Set and . With the trust-region constraint omitted, the SQP increment at iteration solves
Define , . The KKT conditions in are
Therefore, the SQP iteration is the composition
Now we consider the trust region bound , we let and define the neighborhood
We show that, for sufficiently small , the map is a local contraction on the , i.e.,
We omit the proof due to space constraints. By the contraction property of on , the iteration converges to a fixed point at which . Hence, the KKT condition become
Given that
therefore, with muliter , we have
Compared with (28), this system use instead of . For a backward solution with , its stationarity residual in the above system is
The unit-quaternion identity gives
The local KKT inverse estimate therefore yields
References
- [1]Z. Gu, et al., “Humanoid locomotion and manipulation: Current progress and challenges in control, planning, and learning,” IEEE/ASME Trans. Mechatronics, vol. 31, no. 2, pp. 2300–2330, 2026.arxiv.org/abs/2501.02116
- [2]Y. Gu, et al., “Evolution of humanoid locomotion control,” Sci. Robot., vol. 11, no. 117, p. eead3973, 2026.DOI
- [3]M. Gleicher, “Retargetting motion to new characters,” in Proc. 25th Annu. Conf. Comput. Graph. Interactive Techn., 1998, pp. 33–42.DOI
- [4]K.-J. Choi and H.-S. Ko, “Online motion retargetting,” J. Vis. Comput. Animat., vol. 11, no. 5, pp. 223–235, 2000.doi.org/10.1002/1099-1778(200012)11:5<223::aid-vis236>3.0.co;2-5
- [5]J. P. Araújo, Y. Ze, P. Xu, J. Wu, and C. K. Liu, “Retargeting matters: General motion retargeting for humanoid motion tracking,” in Proc. IEEE Int. Conf. Robot. Automat., 2026.arxiv.org/abs/2510.02252
- [6]R. Villegas, D. Ceylan, A. Hertzmаnn, J. Yang, and J. Saito, “Contact-aware retargeting of skinned motion,” in Proc. IEEE/CVF Int. Conf. Comput. Vis., 2021, pp. 9720–9729.arxiv.org/abs/2109.07431
- [7]J. Zhang, Z. Tu, J. Weng, J. Yuan, and B. Du, “A modular neural motion retargeting system decoupling skeleton and shape perception,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 46, no. 10, pp. 6889–6904, 2024.DOI
- [8]L. Yang, et al., “OmniRetarget: Interaction-preserving data generation for humanoid whole-body loco-manipulation and scene interaction,” in Proc. IEEE Int. Conf. Robot. Automat., 2026.arxiv.org/abs/2509.26633
- [9]U. Helmke and J. B. Moore, Optimization and Dynamical Systems, ser. Communications and Control Engineering. Springer London, 1994.
- [10]J. Schropp and I. Singer, “A dynamical systems approach to constrained minimization,” Numer. Funct. Anal. Optim., vol. 21, no. 3–4, pp. 537–551, 2000.DOI
- [11]S. Fiori, “Quasi-geodesic neural learning algorithms over the orthogonal group: A tutorial,” J. Mach. Learn. Res., vol. 6, no. 26, pp. 743–781, 2005.
- [12]E. Haber, L. Rutthotto, E. Holtham, and S.-H. Jun, “Learning across scales—multiscale methods for convolution neural networks,” in Proc. 32nd AAAI Conf. Artif. Intell., 2018.DOI
- [13]C. Lu, Y. Zhou, F. Bao, J. Chen, C. Li, and J. Zhu, “DPM-Solver: A fast ODE solver for diffusion probabilistic model sampling in around 10 steps,” in Adv. Neural Inf. Process. Syst., vol. 35, 2022.arxiv.org/abs/2206.00927
- [14]H. Klein, N. Jaquier, A. Meixner, and T. Asfour, “A Riemannian take on human motion analysis and retargeting,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 2022, pp. 5210–5217.arxiv.org/abs/2208.01372
- [15]———, “On the design of region-avoiding metrics for collision-safe motion generation on Riemannian manifolds,” in Proc. IEEE/RSJ Int. Conf. Intell. Robots Syst., 2023, pp. 2346–2353.arxiv.org/abs/2307.15440
- [16]J. Li, X. Song, S. Wang, H. Huang, and Y. Zhao, “A unified conditional flow for motion generation, editing, and intra-structural retargeting,” arXiv preprint arXiv:2604.13427, 2026.DOI
- [17]W. Kim, T. Li, and S. Ha, “MoReFlow: Motion retargeting learning through unsupervised flow matching,” arXiv preprint arXiv:2509.25600, 2025.arxiv.org/abs/2509.25600
- [18]E. Hairer and C. Lubich, “Numerical solution of ordinary differential equations,” in The Princeton Companion to Applied Mathematics. Princeton University Press, 2015, pp. 293–305.
- [19]E. Hairer, C. Lubich, and G. Wanner, Geometric Numerical Integration: Structure-Preserving Algorithms for Ordinary Differential Equations, 2nd ed., ser. Springer Series in Computational Mathematics. Springer, 2004, vol. 31.
- [20]E. Hairer, “Geometric integration of ordinary differential equations on manifolds,” BIT Numer. Math., vol. 41, no. 5, pp. 996–1007, 2001.DOI
- [21]H. Si and K. Gärtner, “Meshwise piecewise linear complexes by constrained Delaunay tetrahedralizations,” in Proc. 14th Int. Meshing Roundtable. Springer, 2005, pp. 147–163.DOI
- [22]O. Sorkine, D. Cohen-Or, Y. Lipman, M. Alexa, C. Rössl, and H.-P. Seidel, “Laplacian surface editing,” in Proc. Eurographics/ACM SIGGRAPH Symp. Geometry Process., 2004.
- [23]R. Bergmann and R. Herzog, “Intrinsic formulation of KKT conditions and constraint qualifications on smooth manifolds,” SIAM J. Optim., vol. 29, no. 4, pp. 2423–2444, 2019.DOI
- [24]E. Hairer and G. Wanner, Solving Ordinary Differential Equations II: Stiff and Differential-Algebraic Problems, 2nd ed., ser. Springer Series in Computational Mathematics. Springer, 1996, vol. 14.
- [25]V. Solo and Z. Wang, “Convergence of a tangent space numerical scheme for a stochastic differential equation on a sphere,” in Proc. 60th IEEE Conf. Decis. Control, 2021, pp. 5814–5819.DOI
- [26]J. Li, J. Wu, and C. K. Liu, “Object motion guided human motion synthesis,” ACM Trans. Graph., 2023.arxiv.org/abs/2309.16237