Combining sensory modalities and exploratory procedures to improve haptic object recognition in robotics

Higy, Bertrand; Ciliberto, Carlo; Rosasco, Lorenzo; Natale, Lorenzo

doi:10.1109/humanoids.2016.7803263

Cited by 10 publications

(5 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Gorges et al [13] merge a sequence of grasps into a statistical description of the object that is used to classify the objects. In a recent work Higy et al [14] propose a method in which the robot identifies an object by carrying out different exploratory behaviours such as hand closure, and weighing and rotating the object. In their method the authors fuse sensor data from multiple sensors in a hierarchical classifier to differentiate objects.…”

Section: Introductionmentioning

confidence: 99%

Controlled tactile exploration and haptic object recognition

Regoli

Jamali

Metta

et al. 2017

2017 18th International Conference on Advanced Robotics (ICAR)

Self Cite

View full text Add to dashboard Cite

In this paper we propose a novel method for in-hand object recognition. The method is composed of a grasp stabilization controller and two exploratory behaviours to capture the shape and the softness of an object. Grasp stabilization plays an important role in recognizing objects. First, it prevents the object from slipping and facilitates the exploration of the object. Second, reaching a stable and repeatable position adds robustness to the learning algorithm and increases invariance with respect to the way in which the robot grasps the object. The stable poses are estimated using a Gaussian mixture model (GMM). We present experimental results showing that using our method the classifier can successfully distinguish 30 objects. We also compare our method with a benchmark experiment, in which the grasp stabilization is disabled. We show, with statistical significance, that our method outperforms the benchmark method.

show abstract

Section: Introductionmentioning

confidence: 99%

Controlled tactile exploration and haptic object recognition

Regoli

Jamali

Metta

et al. 2017

2017 18th International Conference on Advanced Robotics (ICAR)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Training instances in this scenario could be gathered autonomously by detecting invariances in the data which correspond to physical entities (e.g., coherent motion patterns (Wang and Gupta 2015) or bottom-up saliency cues). Strategies specific to the robotic domain could be devised by integrating multiple sensory modalities (Sinapov et al 2014;Higy et al 2016) and a repertoire of explorative actions (Montesano et al 2008;Fitzpatrick et al 2003;Högman et al 2016;).…”

Section: Discussion and Future Workmentioning

confidence: 99%

“…Clearly, visual perception is only one of the possible sensory modalities equipping modern robots, that can be involved in the object recognition process (see for example (Luo et al 2017;Higy et al 2016)). In addition it has been shown that the physical interaction with the arXiv:1709.09882v2 [cs.RO] 3 Jan 2019 environment can be used to aid perception , demonstrating that there is more than "just" vision to object recognition.…”

Section: Introductionmentioning

confidence: 99%

Are we done with object recognition? The iCub robot’s perspective

Pasquale

Ciliberto

Odone

et al. 2019

Robotics and Autonomous Systems

Self Cite

View full text Add to dashboard Cite

We report on an extensive study of the benefits and limitations of current deep learning approaches to object recognition in robot vision scenarios, introducing a novel dataset used for our investigation. To avoid the biases in currently available datasets, we consider a natural human-robot interaction setting to design a data-acquisition protocol for visual object recognition on the iCub humanoid robot. Analyzing the performance of off-the-shelf models trained off-line on largescale image retrieval datasets, we show the necessity for knowledge transfer. We evaluate different ways in which this last step can be done, and identify the major bottlenecks affecting robotic scenarios. By studying both object categorization and identification problems, we highlight key differences between object recognition in robotics applications and in image retrieval tasks, for which the considered deep learning approaches have been originally designed. In a nutshell, our results confirm the remarkable improvements yield by deep learning in this setting, while pointing to specific open challenges that need be addressed for seamless deployment in robotics.

show abstract

“…Visuo-tactile methods have attained attributes such as elasticity, mass and relational constraints [10] and object pose [11], [12]. Shape has been determined with methods such as tactile glances at discrete points on the object [13], visual and tactile feedback from grasping [14], combining visual and tactile exploratory procedures [15] and visuo-tactile fusion [16].…”

Section: Background and Related Workmentioning

confidence: 99%