SyncHeC: Markerless Spatio-Temporal Hand-Eye Calibration via Differentiable Rendering and Continuous-Time Joint Optimization
SyncHeC: Markerless Spatio-Temporal Hand-Eye Calibration via Differentiable Rendering and Continuous-Time Joint Optimization
This research introduces SyncHeC, a framework for markerless hand-eye calibration that jointly estimates camera geometry and time offsets from continuous robot motion using differentiable rendering. Evaluated on a 7-DoF arm, the method achieves 0.21 cm translation error and 2.7 ms latency error against reference benchmarks, outperforming CM-Calib by 3.7x in translation and 2.3x in latency. The approach supports Eye-to-Hand, Eye-in-Hand, and multi-camera setups across seven arms and four depth sensors, with policy training integration showing an 8.1 percentage point increase in success rates across six task families.
The paper presents SyncHeC, a differentiable rendering framework for markerless hand-eye calibration that achieves sub-centimeter translation and millisecond-level latency errors on a 7-DoF arm, with reported improvements over CM-Calib under matched recording conditions.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
translation error
0.21
View original evidence
ic depth complements mask contours, and an offline planner combines spatial sensitivity with temporal excitation. Alternating updates support Eye-to-Hand, Eye-in-Hand, and multi-camera calibration. On a 7-DoF arm, one $6.5$\,s motion yields $0.21$\,cm translation error against a marker-board reference, $0.34$\,cm touch-test error, and $2.7$\,ms latency error relative to an $87$\,ms reference with $95\%$ uncertainty o
Open source S3touch-test error
0.34
View original evidence
ic depth complements mask contours, and an offline planner combines spatial sensitivity with temporal excitation. Alternating updates support Eye-to-Hand, Eye-in-Hand, and multi-camera calibration. On a 7-DoF arm, one $6.5$\,s motion yields $0.21$\,cm translation error against a marker-board reference, $0.34$\,cm touch-test error, and $2.7$\,ms latency error relative to an $87$\,ms reference with $95\%$ uncertainty o
Open source S3latency error
2.7
View original evidence
ic depth complements mask contours, and an offline planner combines spatial sensitivity with temporal excitation. Alternating updates support Eye-to-Hand, Eye-in-Hand, and multi-camera calibration. On a 7-DoF arm, one $6.5$\,s motion yields $0.21$\,cm translation error against a marker-board reference, $0.34$\,cm touch-test error, and $2.7$\,ms latency error relative to an $87$\,ms reference with $95\%$ uncertainty o
Open source S3translation error improvement factor
3.7
View original evidence
f $\pm1.5$\,ms. Against CM-Calib on matched recordings, translation and nominal latency errors improve by $3.7\times$ and $2.3\times$. Experiments span seven arms and four depth sensors. Compensating the recovered offset during policy training and deployment increases success by $8.1$ percentage points across six families and six tasks ($1{,}800$ matched pairs; paired $95\%$ confidence interval $[6.3,9.9]$ points).
Open source S4latency error improvement factor
2.3
View original evidence
f $\pm1.5$\,ms. Against CM-Calib on matched recordings, translation and nominal latency errors improve by $3.7\times$ and $2.3\times$. Experiments span seven arms and four depth sensors. Compensating the recovered offset during policy training and deployment increases success by $8.1$ percentage points across six families and six tasks ($1{,}800$ matched pairs; paired $95\%$ confidence interval $[6.3,9.9]$ points).
Open source S4
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ic depth complements mask contours, and an offline planner combines spatial sensitivity with temporal excitation. Alternating updates support Eye-to-Hand, Eye-in-Hand, and multi-camera calibration. On a 7-DoF arm, one $6.5$\,s motion yields $0.21$\,cm translation error against a marker-board reference, $0.34$\,cm touch-test error, and $2.7$\,ms latency error relative to an $87$\,ms reference with $95\%$ uncertainty o
Open source S3
f $\pm1.5$\,ms. Against CM-Calib on matched recordings, translation and nominal latency errors improve by $3.7\times$ and $2.3\times$. Experiments span seven arms and four depth sensors. Compensating the recovered offset during policy training and deployment increases success by $8.1$ percentage points across six families and six tasks ($1{,}800$ matched pairs; paired $95\%$ confidence interval $[6.3,9.9]$ points).
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org