Editorial: Tactile Intelligence in Robots
Bibliographic record
Abstract
Tactile intelligence has become increasingly important to the development of intelligent robots capable of dexterous skills close to humans. Recent advances in tactile sensors foster the implementation of artificial tactile sensation in robots, where tactile intelligence plays a key role in translating physical signals from tactile sensors to tactile percepts. Amid many possible approaches to the development of tactile intelligence, it is reasonable to translate the principles of information processing of the human nervous system into artificial robot intelligence. For instance, one may create artificial neural networks that mimic somatosensory nervous systems and integrate them with advanced machine learning techniques in order to have robots gain human-like dexterous skills. This research topic aims to highlight the state-of-the-art researches on the implementation of tactile intelligence in robots inspired by neural mechanisms of tactile information processing. It also emphasizes human studies on tactile perception to provide insights on robot learning for object manipulation tasks.A number of studies contribute to the present research topic by providing the latest remarkable findings in tactile intelligence. One of the technical challenges in the implementation of tactile intelligence in robots lies at the interface to sense external mechanical stimulations with high-fidelity. Opposed to the widely distributed mechanoreceptors inside the human hand, the current tactile sensing technology in robotic hands often suffers from the sparseness of sensing points and the lack of spatial resolution. Sun and Martius attempts to overcome this limitation by inferring tactile stimulations at virtual contacts from a limited number of strain-gauged sensor data using machine learning algorithms (https://www.frontiersin.org/articles/10.3389/fnbot.2019.00051 ). They demonstrates the feasibility of reconstructing the location and force magnitudes at multiple contact points of deformable objects from sparse sensor configuration. They achieve this by leveraging machine learning algorithms for the inference of tactile information, nicely showing how robots gain the ability of inference from limited observations, akin to what human intelligence always does.Another study by Richardson and Kuchenbecker contributing this topic investigates more closely such connections between robot tactile sensing and human tactile perceptual attributes (https://www.frontiersin.org/articles/10.3389/fnbot.2019.00116/full ). They particularly focuses on attribute intensity and perceptual variability in natural human tactile perception, which is not available in the current robot tactile intelligence. They collected haptic adjectives for a number of objects from human subjects as well as robotic tactile sensing data for the same objects. Then, they successfully predict the probability distribution of haptic adjectives from the tactile sensing data of an object using a machinelearning algorithm. The study demonstrates the possibility of modeling both intensity and variability of human tactile perception by tactile intelligence in robots. This finding will contribute to move artificial tactile intelligence closer to human perceptual system. Tactile intelligence will become increasingly important as robots function in more openended environments, where robots should adapt to uncertainty of environmental states. In this regard, Seminara and her colleagues review on sensorimotor coupling in robotic control of hands and fingers with an emphasis on connections between human tactile perception and robotic tactile sensing (https://www.frontiersin.org/articles/10.3389/fnbot.2019.00053/full ). In particular, they highlights robotic behavior, goals and tasks in active haptic exploration. This review offers a comprehensive look over the taxonomy of elements for the closed-loop sensorimotor control of robots and will be of great help to those who seek to design a robot with an ability to interact with various real environments adaptively and intelligently.Closing the loop of motor control with sensory feedbacks can lead to the cognitive embodiment of external actuators in humans. This is especially important to the use of robots as an assistive technology in a daily life. Beckerle and his colleagues provides the critical review on the role of tactile perception of sensory feedbacks in the feeling of embodiment while using assistive robotic systems (https://www.frontiersin.org/articles/10.3389/fnbot.2018.00084/full ). Throughout the rigorous review, they suggest practical solutions to enhance embodiment by optimizing tactile feedback in human-robot interactions. This review will be considerably valuable to those who aim to improve the usability of robotic assistive technology by providing real-time tactile feedbacks to the user.We believe that all the contributions to this topic will broaden our understanding of neural underpinnings of tactile perception, foster the development of robots interacting with the world in a more intelligent way, and open a new venue to integrate multiple disciplines in order to pave way to the next-generation human-robot interactions. We hope that readers will also find in this research topic useful insights and promising outlooks for their own researches.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.013 |
| Meta-epidemiology (narrow) | 0.004 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.003 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.003 | 0.003 |
| Scholarly communication | 0.007 | 0.005 |
| Open science | 0.004 | 0.002 |
| Research integrity | 0.014 | 0.015 |
| Insufficient payload (model declined to judge) | 0.043 | 0.028 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".