SoftArm
From Merlin Systems Wiki
Contents |
Soft Robot Arm
Winners of a DTI SMART Development Award
Completion: March 2006
Project Summary
Robot arms are commonplace in factories, and are used industrially for tasks such as assembly, welding and inspection. These industrial robot arms are typically large, heavy devices suited to hostile environments, but increasingly there is a requirement from society for robotics technology that can be used in service roles to assist people. This project is aimed at developing robotics technology suited to these emerging roles some examples include:
Robot Assisted Surgery.
Personal robotics
Assistance for disabled or elderly – for use by a bedside or mounted on a wheelchair.
Animatronics
Mounted on a vehicle for de-mining applications.
Larger robot systems.
Hazardous environments.
The technical challenges to construct a robot arm suitable for use near people is quite different from an industrial robot arm:
- Industrial robot arms are heavy -> A soft robot arm needs to be as light as possible.
- Industrial robots are precise and rigid and will exert maximum force to reach a target end position (even if a human happens to be in the way). The proposed soft arm will be compliant and will ‘give’ if an unexpected obstacle is encountered. Another good example of where compliance is very useful would be a window cleaning robot, an industrial arm would be very prone to breaking the glass given even a slight disturbance, whereas a compliant arm such as the proposed Soft Robot Arm would be able to flex and accommodate small changes.
Industrial arms are usually operated in large open spaces, and the physical size of the robot is not normally a problem. For many service applications this is not the case and a more compact framework with a similar performance to the human arm is the ideal.
Merlin Systems Corp. Ltd successfully completed a feasibility study for an SRA in 2001. Many of the concepts and technologies behind the SRA are based upon doctoral research carried out at the University of Plymouth. The technology and concepts that comprise the SRA are internationally significant and Merlin Systems Corp. Ltd is internationally recognised as a leading supplier of advanced robotics systems.
As part of the feasibility study we were able to demonstrate the application of novel technology to solve the problems associated with service robotics (described above). A functional prototype of an SRA was demonstrated at the conclusion of the project. Following on from the success of the feasibility study Merlin Systems Corp. Ltd has subsequently commercialised elements of the SRA such as the Humanifom Muscle.
Soft actuator technology
Actuators are the means by which the arm is moved. In traditional industrial robots the actuators are generally servo motors, which are heavy, expensive non-compliant structures. Merlin Systems Corp. Ltd are world leaders in artificial muscle technology.
An air muscle actuator consists of an inner rubber bladder that is constrained by an outer braided mesh such that as the bladder is inflated the mesh causes the actuator length to shorten. As the length shortens the actuator is able to perform work i.e. move/rotate a limb.
The behaviour of a soft actuator of this type is analogous to a human muscle. Soft actuators are only able to perform work as they shorten in length. Therefore, to fully move a limb in 1-axis two soft actuators need to be arranged such that one pushes and the other pulls, this arrangement is known as a co-operative antagonistic pair. The force and length of an actuator is dependant upon the pressure of air contained within the actuator.
As part of the feasibility study an enhanced air muscle was demonstrated (incorporating SMART Valves), we propose to extend this principle to build an integrated Servo Air Muscle (patent pending). The Servo Air Muscle (SAM) will utilise the LEX Sensor, the SMART Valve and a custom control system to produce a uniquely flexible integrated actuator.
A further objective of this phase will be to assess the robustness of the muscle and to test the MTBF (Mean Time Before Failure). Safety critical issues will also be examined.
Mechanical structure
One of the features that makes the human arm so special is its compactness and ability to lift large loads. This is largely due to the unique structure of the shoulder joint, which provides a very wide range of motion in all 3-axis of rotation as well as being able to lift large loads. In engineering terms it is relatively simple to produce a spherical ball joint, with a limited range of motion that can lift reasonable loads, but the key technical challenge is to produce a sensor that is able to operate over a wide range of motion in all three axis. Industrial robot arms today use simple hinge or 1-axis rotational joints using sensors such as rotary potentiometers, this is because a three-axis solution does not currently exist. A normal three axis joint is therefore constructed from three 1-axis rotary joints. Due to physical constraints each 1-axis joint is mounted a set distance from each other resulting in a bulky arrangement, which increases control complexity. As part of the previous SMART feasibility project we demonstrated a new sensor called THOR (THree Axis Rotational sensor), which we propose to develop into a pre-production prototype as part of this project. This sensor can then be incorporated into a specially arranged ball-socket to provide 170 deg. rotation in two-axis and 60 degrees of rotation in the third. This compact arrangement will be similar to a human shoulder joint. This unique approach to arm design also means that the linkages between joints no longer need to be rigid heavy structures but can instead be lightweight structures with some flex. By using the simplified control structure it is possible to build a new type of control system (called “digilog”) to provide a much more capable arm.
Control System
An SRA is only as useful as the control system allows it to be. In a traditional six-axis robot arm there are normally multiple ways of achieving the same position, there are always numerous positions, which result in “deadlock”. The simplified joint structure means that this problem will be substantially reduced. The control system provides the necessary electrical impulses to drive the valves and deliver the air energy into the soft actuators. A prototype control system will be produced to control the SRA and to provide an interface for the end-user to express requirements of the arm e.g. position commands. We aim to produce an “pendant” style interface that will be easy for even a novice user to operate with a little practice.
