1^st Workshop on Proximity Perception in Robotics at IROS 2018

Speaker

Joshua R. Smith

Invited Talk

A unified Sensor for Pre-Touch, Touch and Post-Touch Force Measurement

Abstract

In my talk, I will describe recent efforts to create a single sensor that provides pre-touch proximity sensing, can detect the onset of contact, and provides measurements of force after contact. Moreover I will describe the sensor concept, its implementation and evaluation and example applications. This work is in collaboration with Sidd Srinivasa and Patrick Lancaster. 

Speaker

Matthias Brandl

Industry Talk

Ultra Short Range Radar for Human Machine Interaction (HMI)

Abstract

Radar technology evolved from big bulky devices to detect airplanes to small devices that are used in car and even mobile devices. These devices allow a four-dimensional detection of its surroundings by measuring the distance to a target, the horizontal angle (azimuth ϕ), the vertical angle (elevation ϑ) and the velocity of the target. The high bandwidth of today’s radar systems in combination with the frequency-modulation continuous wave (FMCW) scheme results in a length resolution in the centimeter or even millimeter range. This high distance resolution and the three other target informations (ϕ, ϑ, v) are required for radar algorithms like gesture sensing, occupation sensing, or material detection. 

In robotic applications, radar sensors are an ideal enhancement to other sensors.

• Radar systems require low power and are small and cost-efficient.

• They can detect targets that are several meters away.

• Besides the 3D-detection of a target in the room. It allows velocity measurement in parallel to the position measurement.

• Radar algorithms like occupation sensing allow the detection of humans even when the humans are not moving.

• Another interesting application is that radar even allows distinction of different materials in front of the radar sensors.

• When sensing gestures of human, the radar sensors can be used as a machine interface to control robots and so on.

Speaker

Hubert Zangl

Organizer Talk

Near Field Tomography in Proximity Perception

Abstract

Near Field Tomography like Electrical capacitance tomography (ECT) and Magnetic Induction Tomography (MIT) are methods to non-invasively determine the material distribution within a region of interest to provide cross-sectional images of this distribution, which also allows for detection and classification of objects. While the simplicity and robustness of the sensor elements make them very attractive for proximity perception in robotics, the reconstruction problem is ill-posed in particular due to the constraints on the sensor placement. A stable solution of such problems can only be found by utilization of prior knowledge, which typically involves the use of computationally intensive algorithms that do not meet the timing constraints in robotics. Consequently, light weight approaches have been developed. In the talk results of our work on fast reconstruction methods in Near Field Tomography and applications of the approach in object detection and human machine interaction are discussed. 

Speaker

Gordon Cheng

Invited Talk

Making the Best of Pretouch

Abstract

In this talk, I will highlight some of the activities we are performing on tactile interaction, we will show how we have made the best use of multi-modal sensors. Making use of various modalities, such as proximity to better control and enhance the perceptual capability of the robot working in a dynamic environment. Several examples will be given.

Speaker

Nathan Lepora

Invited Talk

Biomimetic Tactile Sensing and Haptics

Abstract

This talk discusses a novel vibrissal tactile sensor - the TacWhisker array - based on modifying a 3D-printed optical cutaneous (fingertip) tactile sensor – the TacTip. Two versions are considered: a static TacWhisker array analogous to immotile tactile vibrissae (e.g. rodent microvibrissae) and a dynamic TacWhisker array analogous to motile tactile vibrissae (e.g. rodent macrovibrissae). Performance is assessed on an active object localization task. Overall, the TacWhisker arrays give a new class of tactile whiskered robots that benefit from being relatively inexpensive and customizable. Furthermore, the biomimetic basis for the TacWhiskers fits well with building an embodied model of the rodent sensory system for investigating animal perception.

Speaker

Keysuke Koyama

Invited Talk

Reactive Control for High-Speed Grasping Using Optical Proximity Sensors

Abstract

In my talk, I will present about an integrated control of a multiple-degrees-of-freedom hand and arm with a high-speed proximity sensing. The high-speed proximity sensing enables real-time control of grasp configurations such as an arm tip vector, wrist orientation and the initial finger positions before grasping. Since the positions and postures are controlled without contact, there is no danger of damaging the object or the robot. 

In order to realize the control, we designed a high-speed proximity sensor for the fingertips of the hand, developed an integrated control system for the hand and the arm, and proposed a reactive architecture based on high-speed proximity sensing. 

Speaker

Robert Haschke

Organizer Talk

Tactile Servoing Algorithms for Manipulation and Object Exploration

Abstract

While we at CITEC, Bielefeld University are aiming to understand and replicate human grasping and manipulation and thus explicitly consider environmental contacts, our approaches to tactile servoing can also be useful for proximity sensing. In my talk, I will shortly introduce some tactile sensors we developed, I will introduce tactile servoing as a simple, but effective approach to tactile-based grasping and manipulation. Combining visual and tactile feedback into a coherent servoing controller, we can explore, grasp and manipulate unknown objects. Using incipient slip detection, a grasp can be stabilized in an online fashion, much like humans do as well.

Speaker

Björn Hein

Organizer Talk

Sensor Technology and Algorithms for Flexible Proximity and Tactile Perception

Abstract

The presentation will introduce the multi-modal sensors currently developed at KIT-IPR. These sensors provide the ability to sense proximity and tactile events. The same measuring principle is used for both modalities. In combination with the modular architecture a very flexible sensors system can be realized.  A brief  overview will be given regarding prototype applications that were already investigated showing the potential of these kind of sensors and how ML techniques can be beneficially used to extract information out of the given sensor signals.