Anthropomorphic design, natural and adaptive locomotion and human like behavior and performance are some of the intrinsic features that have driven the rapid growth of humanoid robots during the past decade. The body development of such a humanoid platform that has the physical capacity of a human being poses many challenges from the mechatronic point of view. These must be addressed in a methodical and concurrent manner in order to co-ordinate and integrate the various components that form the full and complete mechatronic structure.

• The 'iCub' platform has as its aim the replication of the physical and cognitive abilities of a 3½ year old baby. This "baby" robot will act in a cognitive scenario, performing the tasks useful to learning, interacting with the environment and humans.
• The small (104cm tall), compact size (<23kg and fitting within the volume of a child) and high number (53) of degrees of freedom combined with the OPEN approach for research in cognitive development form fundamental differences from the many excellent humanoids already developed.

Coming next
Work towards the development of the new generation of the 'iCub' robot is already in progress. To enhance the iCub body we are currently investigating different mechanical configurations of compliant actuator components. They will be employed to upgrade and finally turn the 'iCub' robotic platform into the first humanoid robot with a full active/passive compliant body. The new 'iCub' will demonstrate superior robustness and high adaptability which will permit a wider range of safe experimentation and interaction scenarios.

One of the ultimate goals in humanoid robotics is reproducing human‐like systems which follow accurate kinematic models and can exhibit properties such as compliance. On one hand the research community is trying to intrinsically include compliance in humanoid structures, in order to assure safety even in a failure condition or to meet energy consumption demands. On the other hand the majority of existing robotic arms are made following a kinematic model that is a simple approximation of the real kinematics of a human arm. The purpose of this project is to embed both the features mentioned (compliance and accurate kinematics) in an arm which will gain the benefits of a bio-inspired design.

A biologically-inspired kinematics can lead to several benefits in terms of: workspace, dexterity, mobility, singularity avoidance if compared to a typical 7-dof arm. In this work the compliant actuation unit of Fig. 1 will be used to embed compliance in the joints. Another field of interest which can be included in this work is the development of safe-oriented control techniques.

In the arm presented here it is expected that there will be benefits in terms of:

• Safety
• Dexterity
• Workspace
• Mobility
• Singularity avoidance

In order to maximise the dexterity of the iCub's hands whilst achieving a compact design the current hands are under-actuated. Although each hand has just 9 controllable actions they in fact both have 20 joints with multiple joints being powered by a single motor. However, despite being perhaps the most compact dexterous hands developed they are still a larger scale that the remainder of the iCub. Each hand is also extremely expensive to manufacture due to the intricate machining required.

Project Aims
The aims of this project are:

• To produce a hand more appropriately scaled for inclusion on the iCub robot.
• To increase the dexterity of the hand to more closely match a true two year old child's hand.
• To identify a design and/or manufacturing technique to reduce the financial cost of producing the hand.

Hand Design
The hand developed consists of 22 degrees of motion of which 18 are independently actuated and the remaining 4 (distal finger joints) being coupled. The thumb and fingers have 4 degrees of freedom with the exception being the middle finger which does not include lateral motion of the metacarpal. Two additional degree of freedom are included in the wrist and a further joint adds flexibility of the palm. The hand is cable/tendon driven with actuators being located in the forearm.

The material used to form the hand needed to be lightweight and sufficiently strong to enable small parts to be fabricated. Initially Aluminium was chosen but high friction would mean bearings would be required and this would increased complexity and therefore costs. Structural plastics have superior friction performance and were investigated.

Complex components would require CNC machining or moulding which is potentially expensive and so 3D printing was chosen. This allowed components to be produced rapidly and for very little cost. The material used is ABS which has both frictional and strength characteristics appropriate for the task.

Prototype Finger
To assess the production technique and the suitability of the material a single test finger was manufactured. A simple test rig was produced allowing the finger joints to be repeatedly cycled through their range of motion. After more than 1000 cycles of the finger it was inspected to identify any areas of wear, none were found therefore increasing confidence in the proposed materials and construction techniques. This allowed a full design to be produced.

Sensors
Each of the directly actuated joints is fitted with an Austria Microsystems programmable magnetic rotary encoder to measure joint positions. A diametric magnet is placed at the centre of each joint and the encoder is then positioned above this mounted to the proceeding finger link. As the joint moves, so the magnet rotates beneath the sensors and a change in magnetic field direction can be detected.

Palm
The palm is divided into three sections with the central palm attaching to the wrist and the index and middle fingers. A second section of palm links between the central palm and the ring and little fingers whilst a third section mounts to the thumb. Forming the palm from three separate parts means its shape can be changed by adjusting the relative angles between each link. This allows the palm to adopt a flat position or to form a cup shape, this ability is in line with a true hand and is used to aid grasping of some objects.

Cost
One of the main requirements of the hand was that it be low cost. The 3D printing technique means the manufacturing costs are considerably lower than for the current iCub hand.

The cost of the hand breaks down as follows

• Motors €2000
• Sensors €250
• Mechanical components and miscellaneous parts <€1000.
• Overall cost of hand < €3500

Until the last decade it was clear that the main approach was the use of heavy, stiff position/velocity and torque actuation units coupled with rigid non back-drivable transmission mechanisms. It has become increasingly clear that this traditional stiff actuation approach has significant performance limitations related to safety, efficiency and the ability to interact with the environment. To address the latest a wide range of experimental novel compliant actuation systems have been developed during the past fifteen years.

This project focuses on the development of a new compact soft actuation unit intended to be used in multi degree of freedom and small scale robotic systems such as the child humanoid robot "iCub" (www.robotcub.org) or the Biologically inspired human-like arm developed at IIT . The miniaturization of the newly developed high performance unit is achieved with a use of a new rotary spring module based on a novel arrangement of linear springs.

This actuation platform exhibits the following characteristics:

• Series Elastic Actuator (SEA)
• High integration density
• Introduces fixed passive compliance
• Full joint state measurement
• Gearbox position
• Outer link position
• Joint torque
• Stiffness is tuned to a desired value (via active compliance control)
• Use iCub compatible components

Text

Text

Text

Text

Text

Text

Text