Tactile-based full-body control
In this talk, I will explain the rich interactions that have been made possible with a large area e-skin. I will demonstrate both upper-body as well as lower-body robot interaction with challenging environments. In so doing, I show that the performance of robots can improve substantially with e-skin and that robots can be kept from damages to themselves and the environment. Furthermore, I show a new method that can reduce the impact forces of walking robots while at the same time providing them with better balancing ability.
Tactile perception on the iCub humanoid robot
Robots can actively interact with the environment to learn about objects and their properties, using their sensory system. Among the other sensors, the sense of touch is particularly interesting for roboticists, as it has great potentials to advance how robots perceive and manipulate objects.
In this talk, I will revise our work on tactile perception with the iCub robot. I will describe the technology of the iCub tactile system, and how it has been designed to cover the body of the robot and customized to be mounted on the hands. I will also present our work on tactile object manipulation and active perception.
Making sense of the physical world with high-resolution tactile sensing
In this talk, I will introduce the development of a high-resolution tactile sensor GelSight, and how it could help robots understand and interact with the physical world. GelSight is a vision-based tactile sensor that measures the geometry of the contact surface with a spatial resolution of around 25 micrometers, and it also measures the shear forces and torques at the contact surface. With the help of high-resolution information, a robot could easily detect the precise shape and texture of the object surfaces being contacted and therefore recognize them. But it can help robots get more information from contact, such as understanding different physical properties of the objects and assisting manipulation tasks. This talk will cover our previous works on both object property perception and slip detection with GelSight sensors. I will also discuss our recent development in tactile sensor simulation and how we envision the work will help the community with the research in tactile sensing.
Slip-Detection and Tactile-Based Grasping
Slip detection is a major prerequisite for successful object grasping and manipulation: Early detection of incipient slip enables a robot hand to stabilize a grasp and thus not loose the object. For manipulation, the notion of slip is an important feedback signal to monitor the intended motion of an object in hand. Starting from different sensor modalities provided by the broad range of different hardware implementations, this talk will introduce various approaches to slip detection, based on shear-force detection, friction cone estimation, or vibration detection.
SoftBOTS – Soft Biomimetic Optical Tactile Sensing
Reproducing the capabilities of the human sense of touch in machines is an important step in enabling robot manipulation to have the ease of human dexterity. A combination of robotic technologies will be needed, including soft robotics, biomimetics and high-resolution tactile sensing. This combination is considered here as a SoftBOT (Soft Biomimetic Optical Tactile) sensor. Here I review the BRL TacTip as a prototypical example of such a sensor. Topics include the relation between artificial skin morphology and the transduction principles of human touch, the nature and benefits of tactile shear sensing, 3D printing for fabrication and integration into robot hands, the application of AI to tactile perception and control, and the recent step-change in capabilities due to deep learning. This talk consolidates those advances from the past decade to indicate a path for robots to reach human-like dexterity.
Neuromorphic touch for robotics
Neuromorphic sensing takes inspiration from biology to encode sensory signals in a way that captures important information about the external world. Neuromorphic vision sensors are known as event-driven cameras and react only to changes in the light impinging on the sensitive areas. They feature extremely low latency and high temporal resolution coupled with high compression and dynamic range. Lately, a similar concept has been applied to pressure sensors, used as tactile sensors. These are extremely useful in any application that entails the physical interaction of an artificial device with the environment, as robotics and prosthetics. In this talk, we will discuss the principles of neuromorphic event-driven sensing, how they are applied to the tactile domain and how can robotics benefit from those.
The role and Peculiarities of (tactile) Sensing in Soft Robotics
Soft robotics represents a conceptual departure from "hard" robotics in the sense that a) functionality previously specified in controllers and perception algorithms is now realized directly in robot hardware and b), as a result of the hardware's new, extended role, many of the existing approaches to sensing cannot be applied any longer as they interfere with this new functionality. In this presentation, I will discuss the core "philosophical" principles underlying soft robotics and their consequences for sensing and sensing technologies. Just like the functionality of a robot can be distributed across hardware and software, sensing can also be distributed across an implicit portion (achieved by hardware but without specific sensors) and an explicit portion (as in traditional robotics realized through dedicated sensing hardware). The distribution of sensing requirements across explicit and implicit parts greatly decreases the complexity of sensing, but it also reduces the need for sensing to achieve competent control. To achieve this, so-called "computational sensors" are a key concept. These sensors measure relatively generic properties of an object/robot and then extract from the generic sensor information an application-specific signal through computation. I will argue that this greatly simplifies sensing for many applications and can even be beneficial in traditional robotics. To conclude I will argue that sensing in soft robotics differs conceptually from sensing in "hard" robotics as a) it does not always have to be explicit and b) it can better exploit physical properties of the system to facilitate sensing with computational sensors.
Active Touch Sensing in Mammals and Robots
Active sensing is the control of the sensory apparatus so as to maximize information gain in relation to current goals. In biological
systems, active sensing generally involves tight feedback loops between sensor and effector systems, these have been widely studied
in invertebrate and vertebrate model systems including humans. The understanding of active sensing, and of related perceptual
processes (active perception), is also being advanced through the development of computational and physical models. Robots,
in particular, can provide compelling platforms for evaluating hypotheses about the biology of active sensing since they face the
same challenge of operating in the physical world and make transparent the mechanisms required to implement closed-loop active
sensing control. This talk reviews the current understanding of active touch sensing in mammals, focusing on vibrissal (whisker)
touch and human cutaneous (fingertip) touch, and explores how robotic modeling is casting light on some of the principles and
processes underlying biological active touch.
