Learning to Predict Friction and Classify Contact States by Tactile Sensor

Zhou, Xingru; Zheng, Zedong; Zhu, Xiaojun; Liu, Houde; Liang, Bin

doi:10.1109/case48305.2020.9216788

Cited by 6 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A large variety of tactile sensors have been developed in industry and literature, typically trading between resolution, affordability and sensitivity, image-based 3 [24] and magneticbased [39] sensors. Video prediction models have been applied to tactile image prediction using image-based tactile sensors [24,38].…”

Section: Related Workmentioning

confidence: 99%

“…Nonetheless, we chose to use the Xela uSkin magnetic based tactile sensor due to its low comparative cost, its high frequency readings which are essential for control and the extra challenge of analysing non-calibrated Xela readings (absolute value of the Xela sensor readings depends on the contact force and contact geometry). Zhou et al [39] converted the Xela uSkin tactile sensor readings to a visual representation that could be applied to the CDNA architecture. However, there are significant issues with the proposed representation.…”

Section: Related Workmentioning

confidence: 99%

“…First, the resolution of the image reduces the resolution of the tactile readings; Second, the taxel objects cross over producing an impossible problem for the prediction model to interpret. Using this representation, Zhou et al [39] proposed a simplified version of the CDNA model to perform tactile prediction. However, the CDNA inspired model produced poor test scores for the reasons outlined.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Action Conditioned Tactile Prediction: a case study on slip prediction

Mandil¹,

Nazari²,

E³

2022

Preprint

View full text Add to dashboard Cite

Tactile predictive models can be useful across several robotic manipulation tasks, e.g. robotic pushing, robotic grasping, slip avoidance, and in-hand manipulation. However, available tactile prediction models are mostly studied for image-based tactile sensors and there is no comparison study indicating the best performing models. In this paper, we presented two novel data-driven action-conditioned models for predicting tactile signals during real-world physical robot interaction tasks (1) action condition tactile prediction and (2) action conditioned tactile-video prediction models. We use a magnetic-based tactile sensor that is challenging to analyse and test state-of-the-art predictive models and the only existing bespoke tactile prediction model. We compare the performance of these models with those of our proposed models. We perform the comparison study using our novel tactile enabled dataset containing 51,000 tactile frames of a real-world robotic manipulation task with 11 flat-surfaced household objects. Our experimental results demonstrate the superiority of our proposed tactile prediction models in terms of qualitative, quantitative and slip prediction scores.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Action Conditioned Tactile Prediction: a case study on slip prediction

Mandil¹,

Nazari²,

E³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…For example, in [18], discrete pressure maps provided by tactile sensors were employed to predict grasp stability. In contrast, the solution proposed in [19] leveraged a vision-based tactile sensor and the properties of recurrent neural networks to predict the variation trend of the frictional force at the moment of contact, and classify the contact state. Data-driven methods generally provide computationally efficient inference, which may be critical for detecting slippage in time, and overcome the need to model contact.…”

Section: A Related Workmentioning

confidence: 99%

“…However, these approaches are very sensitive to training data, which are still hard to collect consistently. In fact, to date, the most common way of labeling slip data is by doing it manually (sometimes referred as "expert labeling" [20], [19]), relying on human intuition. This process is generally time-consuming for large datasets and often results in imprecise labels, due to errors introduced by the operator.…”

Section: A Related Workmentioning

confidence: 99%

Leveraging distributed contact force measurements for slip detection: a physics-based approach enabled by a data-driven tactile sensor

Griffa¹,

Sferrazza²,

D’Andrea³

2021

Preprint

View full text Add to dashboard Cite

Grasping objects whose physical properties are unknown is still a great challenge in robotics. Most solutions rely entirely on visual data to plan the best grasping strategy. However, to match human abilities and be able to reliably pick and hold unknown objects, the integration of an artificial sense of touch in robotic systems is pivotal. This paper describes a novel model-based slip detection pipeline that can predict possibly failing grasps in real-time and signal a necessary increase in grip force. As such, the slip detector does not rely on manually collected data, but exploits physics to generalize across different tasks. To evaluate the approach, a state-of-the-art vision-based tactile sensor that accurately estimates distributed forces was integrated into a grasping setup composed of a six degreesof-freedom cobot and a two-finger gripper. Results show that the system can reliably predict slip while manipulating objects of different shapes, materials, and weights. The sensor can detect both translational and rotational slip in various scenarios, making it suitable to improve the stability of a grasp.

show abstract