State estimation for a hybrid rigid-continuum space manipulator system

Lorenzo Ticozzi; Patricio A. Vela; Panagiotis Tsiotras

cs.ROOct 27, 2026

State estimation for a hybrid rigid-continuum space manipulator system

Authors: Lorenzo Ticozzi, Patricio A. Vela, Panagiotis Tsiotras

Organizations: School of Aerospace Engineering, Georgia Institute of Technology · School of Electrical and Computer Engineering, Georgia Institute of Technology

Abstract

Spacecraft-mounted robotic manipulators can support long-term operability of orbiting assets by conducting tasks such as on-orbit assembly, repair, upgrade, and refueling. In this context, the integration of continuum/soft robotic mechanisms into space manipulator designs promises to reduce stowage volume and mass, as well as to improve operational safety through the inherent mechanical compliance of soft structures. This work presents a novel state reconstruction pipeline for a hybrid soft-rigid space manipulator, in which the distal link is a tendon-driven continuum robot. The estimation problem is formulated as a maximum a posteriori (MAP) inference problem over the hybrid manipulator's configuration, conditioned on discrete, noisy measurements from joint encoders and a vision-based sensing pipeline. The resulting soft-constrained optimization problem is transcribed as a factor graph, providing a compact representation for fusing heterogeneous measurements with the rigid-soft kinematic model. Furthermore, the adoption of the Geometric Variable Strain (GVS) approach yields a minimum-order parameterization of the configuration, a property expected to facilitate model-based feedback control design in future work. The proposed implementation is validated in simulation on a closed-loop trajectory where, for better realism, position measurements for the continuum link are obtained by means of a simulated visual pipeline leveraging synthetic photorealistic images. The state reconstruction accuracy and computational performance confirm the potential of the proposed framework for hybrid soft-rigid robotic applications in orbit.

Figures & tables

Figure 1: Diagram of a hybrid space manipulator system. The hybrid soft-rigid manipulator is attached to the satellite bus and features two rigid links, actuated by revolute joints, and a tendon-driven continuum link. Frame $\mathcal{I}$ can be interpreted, for instance, as the Hill frame centered on a client satellite.

Figure 2: Factor graph example. Variable nodes are represented as circles, while factor nodes are represented as squares.

Figure 3: Hybrid space manipulator factor graph. Variables are represented by circles (yellow for poses, green for scalars and vectors); factor nodes are represented by squares.

Figure 4: Upper face of the HSMS base with the left and right camera frames, denoted by $\mathcal{C}_{L}$ and $\mathcal{C}_{R}$ , respectively. The camera elevation angle and stereo baseline are denoted by $\alpha$ and $b$ , respectively, with $\alpha=41.6^{\circ}$ and $b=0.4$ m.

Panel of Figure 5 — Figure 5: HSMS configuration along the $60$ -s simulated maneuver, with the end-effector trajectory displayed in white.

Figure 6: Blender render of the left-camera view. The continuum link carries five colored bands at equally-spaced arclength coordinates, which are detected in both views and triangulated to obtain the position measurements $\tilde{\bm{p}}_{i}^{\mathcal{C}_{L}}$ in Eq. ( 25 ).

Figure 7: Color-based stereo matching across the left and right images.

Covariance	Value
$\bm{\Sigma}_{\mathcal{B}}$	$3\cdot\mathrm{blkdiag}(\sigma^{2}_{\mathcal{B},\mathrm{rot}}\bm{I}_{3},\,\sigma^{2}_{\mathcal{B},\mathrm{pos}}\bm{I}_{3})$
$\bm{\Sigma}_{p}$	$9\cdot 10^{6}\bm{I}_{8}$
$\bm{\Sigma}_{\mathcal{J}}^{R}$	$1\cdot 10^{-8}\bm{I}_{6}$
$\Sigma_{q}$	$3\cdot\sigma_{q}^{2}$
$\bm{\Sigma}_{\mathrm{off}}$	$1\cdot 10^{-8}\bm{I}_{6}$
$\bm{\Sigma}_{M}$	$1\cdot 10^{-10}\bm{I}_{6}$

Table 1: Covariance values for the HSMS factors.

	Position [mm]						Orientation [deg]
Frame	mean	% $d$	max	emp	pred	ratio	mean	max	emp	pred	ratio
$\mathcal{B}$	16.13	0.41%	41.39	17.49	30.00	0.583	1.60	4.92	1.73	3.01	0.577
$\mathcal{R}_{1}$	37.03	0.95%	133.48	41.14	70.95	0.580	1.60	4.92	1.73	3.01	0.576
$\mathcal{R}_{2}$	66.52	1.70%	261.70	74.43	128.62	0.579	1.60	4.92	1.73	3.01	0.577
$\mathcal{F}$	80.39	2.06%	307.83	90.22	156.13	0.578	1.60	4.92	1.73	3.01	0.577
$\mathcal{M}$	87.04	2.23%	323.17	97.66	169.20	0.577	1.76	6.13	1.92	3.64	0.527

Table 2: Error statistics of the estimated frame poses with respect to ground truth, expressed in the world frame. The columns mean and max report the trajectory mean and maximum of the error magnitude

\|\bm{e}(t_{k})\|

, with the mean position error additionally expressed as a percentage of the characteristic length of the system,

d=3.905

m, taken as the Cartesian distance between the base frame

\mathcal{B}

and the tip frame

\mathcal{E}

in the home configuration. The columns emp and pred report the empirical and estimator-predicted standard deviations, RMS-combined across axes as

\sqrt{\mathrm{tr}\,\bm{\Sigma}}

; ratio is their quotient.

Panel of Figure 8 — Figure 8: Ground-truth versus estimated pose of frames $\mathcal{F}$ , $\mathcal{M}$ , and $\mathcal{E}$ w.r.t. the global frame $\mathcal{I}$ , with estimator-predicted $\pm 3\sigma$ intervals.

Figure 9: Ground-truth versus estimated pose of the end-effector frame $\mathcal{E}$ w.r.t. the base frame $\mathcal{B}$ .

Timesteps	1,801
Avg solve time	1.230 ms/timestep
Avg LM iterations	7.271
Max solve time	4.462 ms
Max LM iterations	24
Total solve time	2.216 s

Table 3: Computational performance data.

Robot State Estimation

Robot state estimation infers a robot’s position, motion, orientation, and internal configuration from noisy sensors and physical constraints.

See all 200 papers

Factor Graph-Based Continuum Robot State Estimation with Magnus Expansion Priors
Apr 17, 2026Lorenzo Ticozzi, Sabyasachi Dash, Yangkun Liu +2Factor Graph Optimization Robotics
School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA, USA · School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA
Moving-Horizon Estimation and Nonlinear Model Predictive Control of Cable-Driven Soft Manipulators
Jul 27, 2026Lingxiao Xun, Haihong Li, Gang ZhengNonlinear MPC Moving Horizon Estimation
Observability-Informed Optimal Sensor Placement for Soft Robots
Sep 27, 2026Samuel Smocot, James Richard Forbes, Audrey SedalSoft Robotics
Dept. of Mechanical Engineering, McGill University, Montreal, Canada. · Mila – Québec AI Institute, Montreal, Canada.

Apr 17, 2026J/K move · Enter open · S save

Factor Graph-Based Continuum Robot State Estimation with Magnus Expansion Priors

Lorenzo Ticozzi, Sabyasachi Dash, Yangkun Liu, Patricio A. Vela+1

School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA, USA · School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA

Factor Graph Optimization Robotics Robot State Estimation Continuum Robot Shape Estimation

Open paper AskPDF arXiv

Reconstructing the shape of continuum manipulators from sparse, noisy sensor data is a challenging task, owing to the infinite-dimensional nature of such systems. Existing approaches broadly trade off between parametric methods that yield compact state representations but lack probabilistic structure, and Cosserat rod inference on factor graphs, which provides principled uncertainty quantification at the cost of a state dimension that grows with the spatial discretization. This letter combines the strengths of both these approaches by estimating the coefficients of a compact Geometric Variable Strain (GVS) parameterization within a factor graph framework. A novel kinematic factor, derived from the Magnus expansion of the strain field, encodes the closed-form rod geometry as a prior constraint linking the GVS strain coefficients to the backbone pose variables. The resulting formulation yields a compact state vector, is directly amenable to model-based control, and it retains the modularity, probabilistic treatment, and computational efficiency of factor graph inference. The proposed method is evaluated on both simulated and experimental data and compared against a Gaussian Process (GP) regression baseline, demonstrating increased accuracy in partial and sparse measurement scenarios while preserving computational efficiency.

Abstract

Robot State Estimation

Explore similar work

Factor Graph-Based Continuum Robot State Estimation with Magnus Expansion Priors

Moving-Horizon Estimation and Nonlinear Model Predictive Control of Cable-Driven Soft Manipulators

Observability-Informed Optimal Sensor Placement for Soft Robots