CENTRE FOR THEORETICAL NEUROSCIENCE, UNIVERSITY OF WATERLOO

Project: Characterization of Grasping Decisions With a Human-Operated Gripper

People: Rajan Iyengar, Bryan Tripp

 

The lab's goal is to understand the information processing performed by the primate visual cortex. The group develops computational models of the visual cortex, and compares these models to neural recordings and human behavior. The team is particularly interested in the visual guidance of movement, including control of the eyes, hands, and body.

 

The group is developing an intuitive interface for direct human control of the ReFlex Hand, and is using this to study visually guided grasping decisions in humans. 

 

 

GRASP LAB, UNIVERSITY OF PENNSYLVANIA

Project: A Triangle Histogram for Object Classification by Tactile Sensing

People: Mabel M. Zhang, Monroe D. Kennedy III, M. Ani Hsieh, Kostas Daniilidis

 

The lab uses the ReFlex Hand for tactile object recognition and active tactile sensing. Recent development in computer vision and tactile hardware has allowed tactile object recognition approaches to treat dense tactile array inputs as images and to apply standard vision feature descriptors. Such dense sensing arrays are costly. The ReFlex Hand provides affordable and compliant hardware to demonstrate the robustness of our object classification method, which uses sparse touch inputs and does not rely on highly accurate hand proprioception. The team aims to actively determine a minimal set of contact points to touch in order to identify an everyday object.

 

 

HARVARD BIOROBOTICS LAB, HARVARD UNIVERSITY

Project: Tactile-Sensing-Based Grasp Prediction and Refinement

People: Qian Wan, Ryan P. Adams, Robert D. Howe

 

The group's research focuses on using tactile information to enable and improve autonomous grasping and manipulation in unstructured environments.

 

For this project, the ReFlex hand is used for collecting tactile information during real, as opposed to simulated, grasping. The team has collected thousands and thousands of trials to use in our machine learning algorithms, and the hand has been very robust throughout.

 

 

INTELLIGENT AUTONOMOUS SYSTEMS, TU DARMSTADT°

UNIVERSITY OF UTAH ROBOTICS CENTER, UNIVERSITY OF UTAH°°

Project: Learning Robot In-Hand Manipulation with Tactile Features

People: Herke van Hoof°°, Tucker Hermans°, Gerhard Neumann°°, Jan Peters°°

 

Creating autonomous robots that can learn to assist humans in situations of daily life is a fascinating challenge for machine learning. While this aim has been a long-standing vision of artificial intelligence and the cognitive sciences, we have yet to achieve the first step of creating robots that can learn to accomplish many different tasks triggered by environmental context or higher-level instruction. The goal of the robot learning laboratory is the realization of a general approach to motor skill learning, to get closer towards human-like performance in robotics. The team focuses on the solution of fundamental problems in robotics while developing machine-learning methods. 

 

Exact object models are not available for unknown objects. Instead of relying on models, we exploit compliance and tactile feedback to adapt to unknown objects. However, compliant hands and tactile sensors are themselves difficult to model. Therefore,  in-hand manipulation skills are acquired through reinforcement learning.

 

 

INTELLIGENT MOTION LAB, DUKE UNIVERSITY

Project: The Tele-Robotic Intelligent Nursing Assistant (TRINA)

People: Jane Li, Peter Moran, Carrina Dong, Ryan Shaw, Kris Hauser

 

During outbreaks of contagious diseases, healthcare workers are at high risk for infection due to routine interaction with patients, handling of contaminated materials, and challenges associated with safely removing protective gear. This project is developing the Tele-Robotic Intelligent Nursing Assistant (TRINA), a remote-controlled robot to address these challenges. Such robots could perform common nursing duties inside hazardous clinical areas, which could reduce infection risk to healthcare workers by minimizing exposure to contagions and other biohazards.

 

The group is investigating several aspects of the tele-robotic nursing problem, including intuitive input devices, improved contextual awareness, and operator assistance algorithms that automate or partially-automate tedious and error-prone tasks. External collaborators are also studying the use of mobile sensors for providing rich information to the nurse, methods for enhancing the robot's manipulation capabilities, and designing disposable protective coverings.

 

 

JET PROPULSION LABORATORY*

UCLA CENTER FOR VISION, COGNITION, LEARING, AND AUTONOMY**, UCLA 

Project: Human-Robot Collaboration (Hu-Ro-Co)

People: Brandon Rothrock*, Yixin Zhu**, Mark Edmonds**, Siyuan Qi** ,Tianmin Shu**, Steven Holtzen**, Dan Xie**, Song-Chun Zhu**

 

The Center for Vision, Cognition, Learning, and Autonomy (VCLA) is affiliated with the Departments of Statistics and Computer Science at UCLA. The team starts from Computer Vision and expands to other disciplines. The objective is to pursue a unified framework for representation, learning, inference and reasoning, and to build intelligent computer systems for real world applications. The projects span four directions: Vision: image and video parsing, scene understanding, scheduling top-down/bottom-up processes, spatial-temporal-causal and-or graphs; Cognition: functionality, intuitive physics, intentionality, perceptual causality, theory-of-mind, and visual persuasion; Learning: information projection, stochastic grammars, and-or graph learning, and lifelong communicative learning; Autonomy: human robot collaboration, multi-agent task planning, situated dialogue, human value and moral norm.

 

 

MECHATRONICS IN MEDICINE LAB, IMPERIAL COLLEGE LONDON

Project: Characterisation of Compliant Gripper Optimal Strength with Remote Control and Monitoring

People: Mr. Jason Spiliotopoulos, Dr. Riccardo Secoli, Dr. Fangde Liu, Dr. Ferdinando Rodriguez y Baena

 

The lab was set up in 1993 as part of the Department of Mechanical Engineering, to research and develop robotic and mechatronic aids to surgery. The group has developed robotic and mechatronic applications in fields as diverse as urological surgery and training systems, neurosurgery, orthopaedics, high intensity focused us and blood sampling. Outstanding results in the field of robotic-assisted orthopaedic surgery led to the spin-out of a company, Acrobot Ltd, for commercial exploitation of a robot named Sculptor™, an active-constraint system for knee surgery, and WayFinder™, a computer assisted surgical system for minimally invasive hip resurfacing.

 

 

PRACSYS RESEARCH GROUP, RUTGERS UNIVERSITY

Project: Manipulation Planning in Tight Spaces for Logistics Applications

People: Zacharias Psarakis, Hristiyan Kourtev, Zakary Littlefield, Shaojun Zhu, Kostas Bekris

 

The lab works in robot motion planning and manipulation challenges. The focus is on efficiently computing safe and effective motions, while providing performance guarantees. Application domains include industrial and logistics applications, as well as service robotics. The lab is also interested in multi-agent and interactive challenges, where many autonomous agents operate in the same environment or they have to coexist and coordinate with people.

The group uses the ReFlex hand for performing real-world grasping experiments in narrow spaces, such as shelving units used by Amazon Inc.  We have performed a comparative evaluation of the ReFlex hand and a custom-made vacuum gripper in the context of the Amazon Picking Challenge