Hardware-as-Regularization: Enhancing Data Consistency in Imitation Learning via an Adjustable Passive Compliant Leader Arm

Hideyuki Ichiwara; Yushi Kaida; Hirokazu Ishida

cs.ROOct 16, 2026

Hardware-as-Regularization: Enhancing Data Consistency in Imitation Learning via an Adjustable Passive Compliant Leader Arm

Authors: Hideyuki Ichiwara, Yushi Kaida, Hirokazu Ishida

Organizations: R&D Division SB Intuitions Corporation, Japan

Abstract

Low-cost passive leader arms enable scalable data collection for imitation learning, but their lack of restoring forces leaves redundant joints unconstrained, yielding highly inconsistent joint-space trajectories. We address this with a low-cost passive joint module (an adjustable mechanical clutch and a 3D-printed spiral spring) that pulls each joint toward a user-defined neutral posture. This passive restoring force acts as a physical prior equivalent to joint-space $L_2$ regularization (``Hardware-as-Regularization''), improving kinematic consistency with shorter completion time than an unregularized arm and preserved movement smoothness. Experiments show substantial variance reduction over a naive passive arm, trajectory consistency on par with an active virtual spring of the same actuator class, and substantially higher downstream success rates (e.g., 73.3% vs. 36.7% on Block Stacking); a stiffness sweep shows consistency degrading monotonically as stiffness decreases, confirming that mechanical stiffness behaves like a regularization weight.

Figures & tables

Figure 1: Overview of Hardware-as-Regularization. We introduce a purely mechanical joint module that consists only of 3D printable parts, screws, and bearings. By building a redundant teleoperation arm (e.g., GELLO) with this module containing a passive spring and an absolute encoder, we can mechanically regularize joint angles. This physical prior effectively solves the optimization problem $\min_{q}\frac{1}{2}(q-q_{0})^{T}K(q-q_{0})$ subject to the task-space constraint $f(q)=x_{d}$ , allowing users to collect highly consistent imitation learning data.

Metric	Naive GELLO	GELLO w/ v.s.	Proposed
Restoring Force	None	Active (Motor Torque)	Passive (Spiral Spring)
Data Consistency	Low (Null-space drift)	High	High (Physical Prior)
Latency	Zero	High (Control loop)	Zero (Continuous time)
System Stability	Passive (no energy injection)	Conditionally Stable	Passive (no energy injection)
Cost	$\sim$ $200	$>$ $700 (Current controllable motor)	$\sim$ $100

Table 1: Comparison of Leader Arm Architectures for Data Collection

Figure 2: Exploded view of the proposed compliant joint module: a 3D-printed spiral spring for bidirectional compliance, a Hirth coupling secured via set screws for discrete neutral-position adjustment, and a coaxial magnetic encoder for joint-angle measurement.

Figure 3: Overview of the experimental setups. (a) Stiffness measurement with an analogue torque wrench on the proposed module. (b, c) The leader devices: a GELLO with virtual springs and the proposed passive module. (d) The simulation interface showing the current robot pose and the target goal pose (shadow). (e) The physical workspace for imitation learning with a 7-DoF arm, a fixed camera, and a wrist camera.

Figure 4: Results of the stiffness evaluation. (a) Torque versus displacement for the proposed passive spiral spring module (linear fit: 0.0330 Nm/rad). (b) Torque versus displacement for the active virtual spring, showing intermittent mechanical dead zones. (c) Commanded and measured current versus displacement for the virtual spring (linear electrical response: 29.45 mA/rad).

Figure 6: Comparison of individual operator performance across three kinematic consistency metrics: (a) joint standard deviation, (b) Pairwise DTW, and (c) trajectory smoothness (SPARC). Note that User 1 is an expert, while Users 2–5 are novices. The proposed module consistently minimizes null-space drift (a, b) for each user regardless of skill level, while preserving natural smoothness (c).

Task	Metric	Naive GELLO	GELLO w/ v.s.	Proposed
Pick & Place	Pairwise DTW [deg] (Data)	23.23 $\pm$ 11.01	11.71 $\pm$ 5.51	8.32 $\pm$ 3.14
	SPARC [-] (Data)	-2.87 $\pm$ 0.23	-3.23 $\pm$ 0.23	-2.97 $\pm$ 0.25
	Success Rate (95% CI) [%]	53.3 (36.1, 69.8)	96.7 (83.4, 99.4)	96.7 (83.4, 99.4)
Stack	Pairwise DTW [deg] (Data)	16.35 $\pm$ 6.94	10.23 $\pm$ 3.99	8.24 $\pm$ 2.99
	SPARC [-] (Data)	-2.87 $\pm$ 0.21	-3.89 $\pm$ 0.30	-3.19 $\pm$ 0.33
	Success Rate (95% CI) [%]	36.7 (21.9, 54.5)	66.7 (48.8, 80.7)	73.3 (55.6, 85.8)

Table 3: Comparison of collected-data quality and downstream ACT success rates (95% Wilson score intervals;

n=30

rollouts per condition). Naive GELLO trails the proposed module significantly on both tasks (two-sided Fisher exact test), while proposed vs. virtual spring is not significant on either task (see text).

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Joint	Settling [s]	$\zeta$	$\omega_{n}$ [Hz]
J1	$4.92\pm 0.12$	overdamped	n/a
J2	$1.91\pm 0.18$	$0.15\pm 0.01$	$1.61\pm 0.09$
J3	$5.05\pm 0.11$	$0.09\pm 0.03$	$3.24\pm 0.43$
J4	$2.23\pm 0.15$	$0.09\pm 0.02$	$2.59\pm 0.20$
J5	$1.90\pm 1.08$	$0.13\pm 0.11$	$2.13\pm 0.53$
J6	$3.49\pm 1.44$	$0.10\pm 0.07$	$3.19\pm 0.97$

Appendix

Table 5: Free-decay test summary (10 trials).

\zeta

: damping ratio,

\omega_{n}

: natural frequency.

Component	Quantity	Estimated Cost (USD)
Polycarbonate (PC) Filament	1 spool	$45
ABS Filament	1 spool	$15
AS5600 Magnetic Encoder	8 units	$15
TCA9548A I2C Multiplexer	1 unit	$2
Seeed Studio XIAO RP2040	1 unit	$8
6700ZZ Bearing	8 units	$10

Appendix

Table 6: Bill of Materials for the Proposed Leader Arm

Robot Imitation Learning

Robot imitation learning trains robots to perform tasks by reproducing and adapting behaviors shown in demonstrations.

See all 691 papers

Improving Robotic Imitation Learning via Trajectory Standardization
Jun 22, 2026Licheng Yang, Lingfeng Qian, Fei Zheng +4Robotic Manipulation Robot Imitation Learning
State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), Institute of Automation, Chinese Academy of Sciences · D-Robotics
RESTA: Time-Scaling Augmentation for Data-Efficient Robot Imitation Learning
Oct 13, 2026Yunpeng Jiang, Jianshu Hu, Han Fang +3Data-Efficient Robot Learning Robot Imitation Learning
Global College, Shanghai Jiao Tong University, Shanghai, China · Shanghai Institute of Physical Intelligence and Robotics (PAIR), Shanghai, China · Duke Kunshan University, Jiangsu, China
ReActor: Reinforcement Learning for Physics-Aware Motion Retargeting
May 7, 2026David Müller, Agon Serifi, Sammy Christen +3Cross-Embodiment Robot Learning Motion Retargeting
Disney Research, Switzerland

Jun 22, 2026J/K move · Enter open · S save

Improving Robotic Imitation Learning via Trajectory Standardization

Licheng Yang, Lingfeng Qian, Fei Zheng, Yonghao He+3

State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), Institute of Automation, Chinese Academy of Sciences · D-Robotics

Robotic Manipulation Robot Imitation Learning

Open paper AskPDF arXiv

Imitation learning for robotic manipulation relies on large sets of human demonstration trajectories, which are often noisy and temporally irregular due to variable operator speed, intermittent pauses, and inconsistent action density. A common preprocessing strategy is time-uniform downsampling to shorten sequences, but it cannot effectively remove speed-induced non-uniformity or redundant pauses. This mismatch degrades data quality and hinders policy learning. To address this issue, we propose Information-Standardized Trajectory Resampling (ISR), an offline preprocessing method for effective imitation learning. ISR resamples each trajectory by enforcing approximately equal information distance between adjacent points. Specifically, we map trajectories onto an information-modulated Riemannian manifold and perform geodesic-equidistant parameterization. We construct an information-intensity field from velocity and acceleration norms: the velocity term removes small-motion redundancy, while the acceleration term preserves high-curvature and fine-manipulation phases. We evaluate ISR on three real-world manipulation tasks with mainstream imitation learning policies. Compared with the baseline time-uniform 3x downsampling, ISR improves task success rates by about 25%, remains robust across datasets collected from different operators, and reduces both dataset size and training cost. The code and videos are publicly available at https://d-robotics-ai-lab.github.io/isr.page.

Abstract

Robot Imitation Learning

Explore similar work

Improving Robotic Imitation Learning via Trajectory Standardization

RESTA: Time-Scaling Augmentation for Data-Efficient Robot Imitation Learning

ReActor: Reinforcement Learning for Physics-Aware Motion Retargeting