Steerable string - robots which work on the inside

It could soon be the year of the snake, at least in robotics terms, according to Rob Buckingham, Managing Director of OCRobotics.

Historically, robots have been designed to operate on the outside of objects or to work in an environment where there are very few constraints.

But there are situations when working on the inside is essential.

Examples include inspection of nuclear facilities or industrial equipment or a submarine where the hazards or costs prohibit disassembly.

That's where snake-arm robots come in.

Many dangerous, time-consuming and otherwise impossible tasks can be simplified by these machines, which can reach the heart of a particular application without touching or damaging any components on the way.

A snake robot doesn't have wheels or legs, but moves by flexing its joints.

By using its own body as scaffolding, the robot acquires an extensive motion range, including the vertical direction.

Since a snake robot is composed of a number of similar co-operating cells, where each cell is an independent unit, it becomes easy to manufacture, inexpensive and easy to maintain.

In principle, a snake robot can climb by wriggling up a pole, and climb vertical steps by using its own body as scaffolding.

Applications could include clearing mines or finding survivors after natural disasters, always assuming they would also withstand the shock of seeing an artificial snake coming to rescue them! Snake robots are examples of hyper-redundant robots, or robots with a large number of degrees of freedom.

Interestingly, such robots borrow inspiration from animals, in this case from jellyfish, worms and colonies of one-cell animals such as amoebas and volvox.

From a robotics point of view, these life forms have many interesting features.

They consist of many small parts, which are not complicated, but which combined have advanced properties.

The snake is basically a robotic arm, comprising a number of individual segments, with a drive unit controlling one or more of them.

Taken to its ultimate, each individual segment can be controlled independently and has its own specific motors.

A typical five-segment machine would be powered by 15 motors.

The company is using Maxon Motor A-max or EC models with planetary gearheads and encoders.

For a really flexible 20-segment machine, as many as 60 motors would be used, all computer controlled and working in synchronised motion.

One of the prime target applications is servicing jet engines.

The snake arm robot travels down the air path to inspect inaccessible components, removing the need for manual access or dismantling.

Importantly, the snake arm robot can enter the engine and travel through the fan blades while components are still hit, reducing downtime and improving turnaround.

Other applications planned include operating TV cameras, searching vessels and vehicles for drugs, working in boilers or ovens and drilling underground.

In the medical world, the snake arm robot may be used in keyhole surgery - again, the key advantage over current procedures is that the whole device may be controlled, not just the tip.

These snakes need to be long and flexible - ideally like a piece of steerable string.

OCRobotics is considering arms as long as 15m with a diameter of only 80mm.

These are seriously flexible devices which need support from their environment, but they would be ideal for exploring a collapsed building or exploring the drains.

In effect these devices take the advantages of endoscopes and combine them with the motion control of a robot creating a device that is both flexible and can be controlled to follow a path.

Another key area for snake arm robots is for shorter devices that need to carry a significant payload - lets say an arm with a reach of 3m and a payload of 20kg.

Such an arm is ideal when the working environment is not benign or static.

Oil and gas exploration is now being conducted at depths of more than 2000m.

At this depth everything is conducted remotely so examining a wellhead or removing it at the end of its life requires some very capable remote technology.

Snake-arm robots also create the opportunity to explore old and new worlds in a different way.

So recent press about space exploration being conducted by robots is relevant - it's great to have a set of wheels or legs to manoeuvre over rough terrain but arms are very useful for going beyond a just looking mode and getting your hands dirty - picking up bits of rock to take a closer look or to stick them in your pocket.

Clearly there is a whole load of interest in unmanned military operations.

But whilst there is significant investment in unmanned aircraft and subs, the challenges with traversing a desert or forest or beach are considerable.

Solutions to these challenges are available although the next range of vehicles is likely to be aimed at what may appear more straightforward applications - for instance guarding military or industrial facilities.

One thing that is difficult to get across is just how straightforward it is to control these things.

Basically, the operator sees where the snake-arm is going using a tip-mounted camera and changes direction using a joystick - like flying an aircraft.

The clever bit is the software that keeps track of where the rest of the device is and makes sure that it keeps as close as required to the path the camera has traced.

This means it can avoid obstacles and makes it possible to interactively explore a structure - going forwards then backing up creating a tree of paths.

Of course if the path is known offline then motion control becomes much more straightforward.

OCRobotics is based in Bristol and has close links with the aerospace industry and with local universities.

Formed five years ago, the company has recently won two Smart Award from the DTI.

It launched the snake arm technology in January of this year.

The company presented a paper at a prestigious international workshop in Toulouse this October on the issue of robots in human environments.

The aim of the workshop was to gather the world's leading researchers together to discuss applications ranging from surgery to delivering the internal post.

In this environment, OCRobotics is slightly out of place because our focus is commercial not academic - but OCRobotics will be demonstrating a reliable hardware platform that could be used by researchers to take the subject further.

If robots are to find their way into our homes they will certainly need a map and a way of getting around.

They will also need to be able to get out of the way.

One of the key requirements for a robot interacting with people will be that, when they collide, the robot moves first - but doesn't spill the coffee in the process.

Snake-arm robots offer the right sort of capability to achieve this - you can push the middle of the device out of the way whilst leaving the coffee cup in the same place.

We also have plans to put a skin on our arms so that they can detect and react to contact, heat or chemicals.

This is a new technology.

Converting a piece of string into a piece of steerable string is not simple but we have cracked the key technical challenges.

Recent work at OCRobotics has focused on developing a product family by concentrating on new actuators and a quick release mechanism.

The aim is to be able to present a family of arms that can be manufactured in volume and sold at a very competitive price.

Our Demonstrator only has 17 part drawings - the next version will cut that down to 10.

OCRobotics is looking for partners in the UK, Europe, the USA and Japan - companies who have particular requirements for working in difficult to reach places such as inspecting and maintaining aircraft wings or aero-engines.

It would also be interested in talking to companies who make robots or endoscopes.

Snake-arm robots do not compete directly with these products but could be a very interesting addition to the catalogue.

We are also interested in talking to investors who have a vision for developing robots for human environments - from surgery to vacuum cleaning - not ironing, ironing really is very tricky.

Once we have an understanding of what you need to achieve and can agree a specification we will prepare a detailed quotation.

Currently we are working to deliver arms within 5 months of order but this does depend on the exact requirement.

Price also depends on specification and can vary from GBP 50,000 to GBP 500,000.

It may be appropriate to conduct a short feasibility study, costing GBP 10,000-20,000, before building the first system.

The arm is self-supporting.

We can also build longer thinner arms when such an arm can be supported at points along its length.

Our definition of repeatability is the range of deviation in tip position when returning to a predefined position using the same path.

For a 3m arm the repeatability will be approximately 5mm.

In nearly all applications of snake-arm robots there is no requirement for any absolute accuracy since the operator controls the device relative to the environment (just like driving a car or flying an aeroplane).

However, the computer can calculate where any part of the arm is to within a few millimetres.

We have designs to incorporate sensors into the skin of the device.

This will mean that we are able to control contact with the environment.

This is particularly important where there is interaction with people or fragile components.

The physical limitations of the device relate to complexity of the path, the path following accuracy and the payload.

Each arm is made up of a number of segments.

The shape of each segment is controlled by a series of actuator units.

Typically, we use three motors to control the shape of each segment in two dimensions.

The flexibility of the arm is a function of the number of segments and a couple of parameters in the design of each segment.

The number of segments you require will depend on the path and path-following accuracy.

We have designs for a 20-segment (40 degree of freedom) device.

We use VRML models to represent the environment, either importing models from CAD or loading and modifying files within the user interface.

The size and weight constraints on sensors and tools depend on the diameter and length of the arm.

The payload can vary from 10g to 100kg.

In fact some applications require a payload of 500kg.

The size the drive unit is about the size of a hat box (a cylinder of diameter 500mm and height 500mm) and it weighs approximately 50kg (subject to specification).

Back to top