Variable compliance in a robot arm for safety and shape-adaptable grasping

Adaptable Grasping

Research aims

The goal of this research project is to design and build a robotic arm apt to be used around humans in changing “real world” conditions. For this reason, features such as safety and ability to interact with variable shape objects in and unstructured environment are being addressed.

Structures that are able to go from soft to rigid and vice-versa (variable compliant ones) are exploited in this work in order to design a robot manipulator for human-robot interaction in the context of service robotics. Variable compliance is exploited in a soft-robotics application in the gripper, inspired to hydrostatic skeletons, invertebrates (e.g. snails, earthworms) that are able to vary their internal pressure to transfer forces rather than relying on bones. A smart joint at the base of the arm ensures that the manipulator can passively go from a rigid behaviour in normal operational times to a compliant one in case of collision. In fact, in case of a collision between the human and the robot arm, the latter should give way to the human avoiding a potential hazard. However, normally it is important to maintain a rigid manipulator structure in order to achieve positioning accuracy and simple control.

The whole project seeks to build most of the intelligence in the physical structure obtaining a robotic arm that is inherently safe and a gripper intrinsically able to adapt its shape without relying on active control methods.

In the video, the hydrostatic skeletons inspired gripper is able to grasp a cup and hold it. Initially, the gripper is very soft, allowing it to mould around the object, but as the grasping progresses it passively becomes more rigid which permits a stable holding of the cup.

INTRO - Initial Training Network

This work is part of the INteractive RObotics research network, also known as the INTRO project, a 4-year EU funded Initial Training Network (ITN) in the Marie-Curie People Programme (FP7) that aims to create a new generation of robotic researchers with a broad understanding of the research and technologies needed to build intelligent robots that function in close interaction with humans in unstructured, changing “real word” conditions. Despite intensive R&D efforts in robotics, autonomous robots can still not perform reliably in unknown, dynamic, and populated environments. Current robotic systems perform well only when all conditions are known and well defined. This drawback currently limits the commercial application of robots. To expand robotics applications into real-world environments, the robots must be constructed for a large variety of tasks and be able to adapt continuously to new and changing working conditions. The robot must be delivered with advanced capabilities to learn new tasks and new working conditions from its new user in its new environment. Safety is a critical issue that must be realised before a robot enters everyday human populated environments. The robot must avoid hurting any human and should also prevent damage to itself if and when it receives incorrect human advice. To advance robotics into new domains, such as consumer, service, home and entertainment applications, varying forms of interaction between humans and robots are required; it is these interactions—under the general term of Interactive Robotics—that constitute the overall research theme of the INTRO proposal. The specific research objectives are to improve robots’ capabilities by combining human and robot experience and to provide robots with cognitive and multi-modal interaction abilities.

Back to Safe HRI

Theme Leader

Project Supervisors


Maria Elena Giannaccini
Bristol Robotics Laboratory
University of the West of England
Coldharbour Lane 
Bristol, BS16 1QD
Telephone: +44 (0)117 32 83394

Page last updated 12 May 2016