17
Nature Inspired Robotics Richard Owen (MSc. Advanced Computer Science), Victoria Cope (MSc. Intelligent Systems Engineering), & Pawel Czyszczon (MSc. Intelligent Systems Engineering) University of Birmingham, School of Computer Science

Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Embed Size (px)

Citation preview

Page 1: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Nature Inspired Robotics

Richard Owen (MSc. Advanced Computer Science), Victoria Cope (MSc. Intelligent Systems Engineering), & Pawel

Czyszczon (MSc. Intelligent Systems Engineering)

University of Birmingham, School of Computer Science

Page 2: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Introduction

The standard, clichéd, view of modern robotics---as held by those lay to the topic---will involve a human-like bipedal machine that looks, and, to a certain extent, behaves, like a person. This view shows the classic nature inspired aspect of robotics, but current research in the area has more scope, and more substance. Inspiration is being taken from nature in order to solve many different problems in many different application areas, and it is mainly these projects that are the subject of this paper. However, before such projects are discussed in detail, it is necessary to put robotics research, as it pushes on into the 21st Century, into context.

Modern Robotics Applications

Possibly the sole reason, and certainly the most prominent, for producing robots (in certain cases), is that they undertake tasks that people are either unwilling to do, or unable to do. It is clear that when robots are assigned to such tasks, many will come into contact with regular members of society. Both their acceptance here, and their competence, is vital for their future. In an article relating to application and success of robots, [1] casts a critical eye on two robots in particular: Roomba, a robotic vacuum cleaner, and TUG, a medical supplies transporter.

Roomba is a small, disc-shaped vacuuming robot, which navigates and cleans floors unaided, detects the presence of dirt, and keeps its batteries charged. It is manufactured by iRobot Corp of Massachusetts, and is commercially available. According to numerous studies cited in the article, Roomba surpassed expectations in the majority of cases, and many of those involved in the trial elected to keep the robot. In an online article, Fox News go as far as saying that Roomba owners have become ‘deeply attached’ to their robots, and suggest that this bodes very well for the future of robotics in society [2]. While this product seems to be a success, others invoke more mixed responses.

TUG is a small to medium sized robot which is used to deliver medical supplies among wards in hospitals. The robot is championed briefly by an article at [3] listing its capabilities, and a white paper by TUGs creators, Aethon. It is capable of carrying well over 200 kilograms, navigates intelligently, and can communicate with staff, and other TUGs and elevators through wireless networking. As a side note, the robot was given this application for financial reasons; it may save hospitals as much as $240,000 over a five year period. As has been suggested, not all staff took to TUG. Some thought it was very useful and competent, while others complained that it was overly attention-seeking. It was claimed that the robot would interrupt them at inappropriate moments, and that it did not change its volume in loud environments, making it inaudible. One quote, found in New Scientist and taken from a participant in a TUG trial, sums its shortcomings up nicely:

“I'm on the phone! If you say `robot has arrived' one more time I'm going to kick you in your camera.”

Aethon are currently developing a new model of TUG which will hopefully overcome some of these issues, though the ability to engage a member of staff appropriately may prove a considerable challenge. As ‘New Scientist’ are keen to emphasize, however, these abilities---irrespective of the challenges they impose---are key in sealing the acceptance of robotics in society. It is encouraging to see such trials and improvements being undertaken within robotics, though there are also areas of robotics which, due to added complexity of tasks, are still somewhat theoretical.

Smarter Robotics

In an article from the journal Mechanical Engineering, [4], entitled ‘Toward a Smarter Bot’, the author outlines the challenges that are currently faced by researchers into intelligent robotics. These are, in no particular order, providing robots with the ability to:-

• Use its powers of perception to notice when a problem solving strategy is required

Page 3: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

• Select the correct strategy for an appropriate problem

• To otherwise act in response to the needs of those around it

...without the aid of human intervention.

The article suggests that while these challenges are being met in part, there is still considerable work to be done in putting the pieces together. A team of European researchers is attempting to do just that, being led by Paolo Arena: an electrical engineering professor at the Universita da Catania in Italy. They have produced a software architecture they believe can be used to create more intelligent robots [4]. The software aims to co-ordinate a robots visual, audio, and tactile sensors to produce input patterns that will guide the robots responses. It is claimed by the researchers that the architecture “is a starting step towards emulating the essential perception-action of living beings”. One possible application of such an intelligent robot, as put forward in the article, would be to locate and help people trapped in a collapsed building. The challenges here would be to firstly navigate the locally altered and unstable terrain, perceive the positions of those trapped, and understand how to act accordingly.

Robotic Learning

In a separate paper in Mechanical Engineering, [5], attention is brought to the growing complexity of tasks demanded of robots, and the need to teach the robots how to perform these tasks. A group of Stanford University researchers are developing impressively competent autonomous helicopters that learn in a nature inspired manner: through emulation. It is claimed that these systems are not only capable of managing an inherently unstable system and flying safely, but also performing various acrobatic stunts better than their teachers. This was done by allowing an Artificial Intelligence (A.I) routine to analyse human controlled flights and develop strategies to emulate them. The helicopter will then carry out this strategy by use of sensors that measure its position, direction, velocity, acceleration, orientation and spin 20 times per second, and adjust these parameters accordingly. The main aim of the project at Stanford is to create autonomous helicopters that are capable of searching for specific objects, such as landmines, over a wide area, or mapping the spread of a forest fire. The current results of the research certainly seem to show a step in the right direction, especially the ability to learn to perform tasks as required.

While this is clearly impressive, and shows promise for the future of teaching robots, it is also still human guided. It will quickly become clear that robotics researchers are optimistic as to their successes, but still have extremely complex challenges to overcome, as outlined in section 1.1 above.

Robotic Evolution

An attractive concept in Artificial Intelligence generally involves the gradual evolution of desired behaviours. One of the main reasons for this attractiveness is the simplicity of the technique; all that is required is a measure of error, and a crude natural selection mechanism. The ‘teaching’ of Artificial Neural Networks (ANNs) used to control many robots is already extremely complex. It seems logical, then, that as complexity reaches certain thresholds, simpler techniques should be given an opportunity to develop more advanced levels of behaviour. This area is doubly nature inspired. Firstly there is the desire to provide robots with natural, high level cognitive abilities as seen in humans, and secondly, the technique that is used to do this is nature-inspired: evolution.

In an intriguing paper, Mitri, Floreaon, and Keller have set about researching this precise area [6]. Their research is on the topic of evolutionary communicative behaviour amongst robots, and produces some interesting findings. It is widely regarded that animals learn by taking cues from other animals, and from trial and error. The more successful animals in a population will survive to breed, and intelligent behaviours will grow as this process continues. It is clear to see that should this natural process be simulated, the potential for intelligent behaviours to emerge amongst robots exists.

Page 4: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Furthermore, in a simulated selective environment, the objective function may focus upon literally any behavioural trait (or combination of traits), and this is at the discretion of the programmer.

In the vase of the paper mentioned above, robots would be selected for breeding based on their ability to find and maintain good proximity to a food source. It should be noted that initially the robots were programmed to ‘signal’ other robots when a source of food or poison was found. Robots were also programmed to display light at short intervals. In addition to this, the robots' environment progressively was engineered throughout the experiment so as to necessitate competition among them. Each of the robots behaviour was to be analysed by looking at each of the three following factors:-

• The frequency of blue light in different areas of the arena

• What inferences related to food location could be drawn from blue light

• How the robots responded to blue light [6].

The robots were placed in the arena, (of size 3m x 3m), and allowed to act for 1,200 time units each. Their behaviours were controlled by ANNs made up of 33 randomly assigned “genes”. After these time units, the robots were assigned points each time they were near a food source and deducted points for being near the poison. The robots with higher points were selected, recombined with other selected robots and mutated in order to produce new ANNs which were used in the next generation. The experiment was continued for 500 generations, with 10 robots in each population, and 100 populations were used. Genes were spread amongst populations; robots were not restricted to breeding within their own ‘10 member’ populations. The experiment was repeated independently on 20 occasions.

As the cold evolutionary system that drives the behaviour of these robots may take on complex dynamics, (evolution is effectively a particular instance of mathematical game theory), it is important to attempt to work out what the process is selecting for. It is also helpful to attempt to infer the advantages that certain individuals may have over each other. During the first experiment, little competition was instigated among the robots, and they quickly evolved a co-operating scheme. The blue light was interpreted to mean that a food source was near the light, and so following the light meant finding the food source.

However, when selection pressure increased, natural selection began to favour robots whose behaviour was selfish rather than co-operative. The co-operative robots would inadvertently give themselves competition and decrease their chances of survival. Robots that did not signal in the presence of food typically had less competition and survived. As a consequence, after the first 52 generations, the robots learned to hold important information from each other. This was done in order to suit their own needs. After the 500 generation budget had been used up, some the robots still held the inference that blue light indicates food, and other robots were emitting light randomly.

This experiment is certainly important when it comes to understanding behavioural traits and problem solving within robotics, although there is clearly more work to be done in order to create intelligent robots. The research shows very interesting results, and highlights a new method for studying evolution, but does not allow creation of intelligent robots. For example, it would not be possible to use this strategy to teach a robot how to navigate through a house; consumers would be unwilling or unable to buy dozens of robots when one is all that is wanted, and even less willing to let them loose in the home with undefined behaviour. Nevertheless, it is still a step in the right direction, and more complex evolutionary environments may lead to more sophisticated intelligent behaviour.

Page 5: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Limitations of Modern Robotics

The following section concerns limitations in terms of movement, and especially navigation over varying terrain with respect to modern robotics projects. Although creating a bipedal robot that has the co-ordination levels to walk is very challenging, there are many projects that have achieved this; examples are Spring Flamingo (found in [7]) and Flame (found in [8]).

Navigating Sloped Terrain

In the paper discussing the ‘Spring Flamingo’ robot---[7]---the importance of rough terrain locomotion is re-iterated, and the research is centered upon designing a robot capable of navigating across a sloped area of terrain. It is important to note here that although the research was successful, there were a number of assumptions made about the specific area of terrain. There were that no slipping occurred during the testing, the slope caused no variation of the robots gait, and that the terrain sloped only in one direction. The slope was specified as upwards, at an angle parallel to the robots single direction of movement. As the mechanics of human movement are extremely complex, it is logical to assume that emulating these mechanics effectively in robotics is at least as complex. Indeed, when studying diagrams in [7], it is quite obvious that the project is somewhat removed from the mechanics of human walking. From its findings, it is quite clear that the work presented is an advancement, but has some way to go to be considered groundbreaking. In the field of bipedal robotics, though, there is research that seems to be moving in the right direction.

More Natural Movement

The development of robots ‘Flame’ and ‘Tulip’, as discussed in [8], goes a certain way towards simulating the dynamic movements of human walking more sincerely. In their paper, they pioneer the technique of Limit Cycle Walking, in which balance and centre of mass are used to drive and direct motion to a certain degree. The research, though, is paradoxical, as it aims to build a robot that moves in a similar manner to humans, in order to (in part) provide a vehicle to study how humans move. In fact, in order to realise how complex some of the tasks that humans perform are, it is sometimes necessary to attempt to automate them [9]. Flame, then, moves by effectively falling forwards, but in a controlled, balanced manner. This simulates human walking to a degree, and gives the robot a more fluid, efficient gait [9]. In this regard, the tone of Flames creator in assessing the robot is both optimistic and subdued by further challenges.

The main positive aspect of the analysis is that a robot has been designed that is more versatile and energy-effective than any other; Limit Cycle Walking clearly has impressive potential. Flame is also capable of navigating across surfaces that have step-downs as deep as 8mm in them. However, this brings us back to the problem of uneven terrain: Flame cannot walk over rough terrains, up or down stairs, or over significantly sloped surfaces.

The Need for Nature Inspired Robotics

The limitation here is that the robots mentioned can only navigate across certain pre-defined flat surfaces; surfaces known to the researchers---and available for scrutinization--when the project is being designed. Naturally, should a robot of this type be capable of applications such as exploration or performing household tasks, it must be suited to navigating undefined terrain successfully. Such a task has not yet been achieved, and so this presents a major obstacle for the advancement of modern robotics. Owing to the fact that this task is too difficult for robotics technology in its current state, researchers have, again, looked towards nature in finding inspiration for solving this problem. There are many different animals that have evolved strategies in order to overcome the problems being faced by researchers, giving rise to more and more nature inspired robotics projects.

However, before we delve into the rich, diverse area of nature inspired robotics research, it is important that we make a distinction between ‘nature inspired’ designs, and those that simply

Page 6: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

copy ideas from nature. Perhaps the most prominent lesson on that which truly embodies nature inspired design is the story that lies behind the ‘flight paradigm’ [11]. It is often considered in conventional A.I. that simply imitating the brain will not produce true artificial intelligence; we will have to solve this problem ourselves. This is exactly what happened in early classical avionics research; directly copying nature’s flight methods did not work in a feasible manner, and the avionics methods used today were drawn up from scratch.

Nature Inspired Robotics

‘The idea of building machines that emulate features of animals that we see around us has a long history. Leonardo da Vinci’s drawings of machines that fly like birds are one familiar example.’[12]

Figure 1. Leonardo da Vinci flying machine based on the structure of bird wings.[13]

Although the earlier attempts at flight were through machines such as da Vinci’s, in the 20th century classical engineering produced a more satisfactory design, which is now known as the aeroplane. Although these crafts are not considered directly nature inspired, they do work on air pressure in the same way bird wings do. Added to this, inspiration is being taken from nature in optimizing aeroplanes and their performance [14]. The point here is as mentioned above: blindly imitating nature is not always the answer to problems, though it is often very useful.

‘Biological organisms have evolved to perform and survive in a world characterized by rapid changes, high uncertainty, indefinite richness, and limited availability of information. Industrial robots, in contrast, operate in highly controlled environments with no or very little uncertainty.’ [15] There are many desirable properties of biological organisms including adaptability, robustness, versatility and agility that make nature inspired robotics attractive.

Page 7: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Figure 2. A typical industrial robot.

General robots can perform very well in constrained environments, where change is very unlikely. They can carry out tasks as long as the robot’s surroundings are as expected however they find adaptability to environment changes very difficult, if not impossible.

There are many situations where a robot that can perform well in a changing environment would be useful. These situations include war zones, disaster zones and space exploration. Robots could be used for investigating the area of a natural disaster, looking for survivors and evaluating whether the area is safe or not for people to go in and help. Current robotic constraints make traversing uneven terrains, where the environment could be altering all the time, with people walking around and buildings falling over very difficult.

Why use nature inspired design?

By using nature inspired techniques we can create robots that can adapt to change, whilst still being robust and agile in their environment.

There are many reasons as to why using nature as an inspiration can help robotics. Nature offers engineers new design concepts and ideas, as well as providing existence proofs as to what robotics might be able to achieve. Biology may also provide ideas to make robot behaviour more successful and adaptive because nature’s creatures are well adapted and evolved to thrive in the natural world, so basing robot morphology, locomotion or control on nature should also produce something which can thrive in the natural world. Also, creating robots that are based on a creatures design either the physical appearance or neural control can help biologists to understand the creature better and gain new insights into how it functions.

What can nature do?

Nature’s creatures are capable of many complex tasks. These tasks include:

• On Land Walking, Jumping, Running, Climbing

• In the water Swimming, Diving, Leaping from water

• In the air Flying, Hovering, Gliding, Diving

• Anywhere Lifting, Pulling, Grasping

Page 8: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Evolution has allowed animals to develop to perform and perfect certain tasks very skilfully, as well as being able to perform other tasks and adapt to changes in their environment. Animals have been optimised for different things including speed, lifting and jumping. A few examples are shown below:

• Speed - Cheetah, Falcon• Lifting - Elephant, Rhinoceros Beetle• Jumping - Kangaroo, Flea, Puma• Gliding - Flying Squirrel

Figure 3. A cheetah, two elephants and rhinoceros beetle, all of which have been optimised to perform a particular task, as detailed above.

Nature’s creatures possess many desirable qualities for robots; some of which are shown below:

• Stability• High manoeuvrability• Ability to function in varied environments• Movement in a combination of land, water and air• Land - walking, running, jumping, climbing• Air - flying, gliding, hovering• Ability to carry objects many times their own weight

How can you use nature as an inspiration to robotics?

Inspiration can be taken from nature in a few different ways. You don’t necessarily need to copy the physical appearance or physical design of the robot. You can use the way the animal moves (locomotion) the biological structure of the animal (materials and geometry) or the biological control (sensors, processing and neural model)

There are many robots around that take inspiration from nature, either using the physical structure of the animal or the neural model of the animal. A few brief examples are mentioned below.

Stickybot

Figure4. Stickybot based on a gecko.

Page 9: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

‘Stickybot has specially designed footpads, inspired by the surface of a gecko’s foot, that allow it to climb smooth surfaces.’[16] The researchers wanted to create a wall climbing robot, and they realised early on that making the robots feet sticky would cause issues when detaching the feet from objects. After finding out how geckos climb objects, using tiny hairs on their feet, the robot designers decided to implement this into their design.

Barbara Webb’s cricket

Barbara Webb’s background is in Psychology, with a PhD in artificial intelligence. She is currently a lecturer at the University of Edinburgh. Barbara Webb and her team have created a robot based on the neural model of a female cricket. [17][18][19][20]

Figure 5. A real cricket and Barbara Webb’s robot cricket.

As you can see in figure 5, Barbara Webb’s cricket does not have the same physical appearance of a cricket.

What does the robot cricket do?Female crickets locate male crickets by following the chirps made by the male cricket. The male crickets chirp has a typical frequency of around 4.7kH and around 2 to 3 chirps per second.

Whegs are used for movement, mainly because there is no design that relates to that of the cricket. However, whegs are related to the way in which cockroaches move, using 6 legs and similar gait to that of the cockroach.

The ears, which are microphones, are mounted on the front of the robot, allowing them to pivot with the front whegs. This mimics the way crickets ears move during turning due to crickets ear’s being located on their front legs. The two microphones have a separation of 18mm, and the input from each is delayed and then subtracted from the other. This effectively performs a phase comparison and provides directional information that is frequency dependent. The separation and delays are tuned to make the directional output best for the typical carrier frequency of cricket song – 4.7kH.

It has a neural network closely based on the cricket’s neurophysiology, which determines the response to a sound signal and this is capable of running in real time.

Because the robot’s neural model was based on the neurophysiology of a cricket, certain things had to be taken into consideration. Things such as how fast the robot can move and how easily it can manoeuvre had to be compared to that of crickets, otherwise comparing the results from this robot with that of a female cricket would be useless. Also behavioural rules had to be implemented, as well

Page 10: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

as suitable noise built in. Obviously crickets don’t have loud motors powering their legs, but robots do, and this had to be taken into consideration when designing the robot cricket.

ResultsThe robot was run from three different starting locations: 8 meters straight ahead, 3.5 meters to the left, and 4.5 meters to the right of the speaker. 10 trials were run from each location.

The main contributor of noise to the signal was the robot’s own motors. Cricket’s ears are actually on their forelegs, and it has been shown that during walking, there is substantial interference in the neural encoding of calling songs. Typically, crickets will stop quite frequently during auditory tracking, although this depends on several other factors such as the light level. However, it has also been shown that they are capable of tracking in an experimental paradigm where sound is switched off whenever they stop moving, so stopping cannot be their only strategy for dealing with self-induced noise. It is possible that they may use some kind of filtering for predictable sounds.

The distance over which phonotaxis was performed was more limited than originally hoped, although still comparable to distances travelled by crickets. The robot was able to cope with sound distortion, including interference from its own motors, and with uneven and slippery terrain.

Nature Inspired Robotics – Application Areas

This material shows an example of nature inspired design. Application area of this design may be classified as “Engineering” area. The problem classification is “Engineering Design and Optimization”. This design is based on observing nature. When thinking of sources of inspiration for this topic, one of the first things that come to mind are insects, especially ants.

Insects, in general, are agile creatures capable of navigating uneven and difficult terrain with ease. The leaf-cutter ants (Atta), specifically, are agile, social insects capable of navigating uneven and difficult terrain, manipulating objects in their environment, broadcasting general event messages to other leaf-cutter ants, performing collective tasks, and operating in cooperative manners with others of its kind. With limited individual abilities, ants perform greater tasks by working cooperatively with one another. These traits are desirable in a mobile robot. However, no robots have been developed that encompass all of these capabilities. As such, this research developed the Biologically-Inspired Legged-Locomotion Ant prototype (BILL-ANT-p) to fill the void.

Biological creatures have muscles which are much stronger for their size than comparable mechanical actuators, body tissue is stronger and lighter in weight than artificial materials would be, and (in the case of ants) they are quite tiny.

The goal of this design is to develop a robot that is power and control autonomous; capable of navigating uneven terrain, manipulating objects within the environment, working together cooperatively, and employment of compliance with the environment and other robots; very strong for its size; and is relatively inexpensive compared to other similar robots.

Body

Page 11: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Figure 6. Example of nature inspired design. Hexapod body inspired by ant body. (Taken from http://www.engineering.case.edu/emae/Research_Areas February 2010)

To aid in deciding which features are to be implemented in an artificial robot, an analysis of the physical and social characteristics of the ant was performed. They use six legs to move with agility over difficult terrain, so the robot must be capable of stable walking in any direction on uneven terrain. They have mandibles that can chew and carry food.

Legged robots have the ability to move over uneven and discontinuous terrain with more agility than wheeled or tracked vehicles, within the legged robot community, only robots with at least three degrees-of-freedom per leg are capable of strafing movements such as crabbing (1- and 2-DOF legs can only move in one and two dimensions, respectively; To move in three dimensions, at least three degrees-of-freedom are required).

There are six 3-DOF legs on the BILL-Ant-p robot. Three degrees-of-freedom were chosen as that is the minimum number which allows strafing; a desired trait for the robot to enable more agile movements. Additional DOF would have been redundant for basic walking and would have required a greater amount of complexity, power, and processing to control. Each leg consists of three joints and four segments (Fig.7). The first joint is the body-coxa (BC) joint, which swings the coxa forward and rearward in the body’s dorsal plane. Next is the coxa-femur (CF) joint, which raises and lowers the femur in the leg-based medial plane. Finally, the femur-tibia (FT) joint raises and lowers the femur and attached foot in the leg-based medial plane.

Page 12: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Figure 7. Top-view body layout comparison of ant and the BILL-Ant-p robot. (Taken fro m http://www.engineering.case.edu/emae/Research_Areas February 2010 )

Figure 8. Front left leg attached to the body. (Taken from http://www.engineering.case.edu/emae/Research_Areas February 2010)

Legs Movements

In the mid 1970’s Dr. Holk Cruse began research with stick insects to investigate nervous system feedback mechanisms that control leg movement. By the mid 1980’s Dr. Cruse and others were observing leg movements and formulating mechanisms that cause contra lateral and ipsilateral adjacent legs to influence one another. Figure 4 describes those mechanisms and shows the influence connectivity. These interactions between legs adjust the position of the PEP due to the position and status (stance phase or swing phase) of adjacent legs, causing the stance phase to be elongated or shortened.

Walking/strafing and rotating employ Cruse Mechanisms 1, 2, and 5 to produce varying gaits throughout the range of speeds. It was determined through previous research that Mechanisms 3, 4, and 6 were not necessary to create a full range of speed-dependent gaits.

Page 13: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Figure 9. Cruse’s basic rule set for stick insect walking (Schmitz 1998). (Taken from http://www.engineering.case.edu/emae/Research_Areas February 2010 )

Using basic trigonometry, the x- and y-components of foot movement for each iteration of a walking cycle are determined based on the strafing angle. Strafing movements allow the robot to move in one direction while facing another. The most extreme example of this is crabbing, where the robot faces forward, but moves sideways. Walking is considered as strafing with a heading of 0º(forward).

Figure 10. Walking (left), strafing at 330º (centre), and CCW rotating (right) foot paths (arrows indicate foot swing direction). (Taken from http://www.engineering.case.edu/emae/Research_Areas February 2010)

AquaPenguin a biomechatronic overall concept.

Page 14: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Figure 11. AquaPenguin technology-bearer for the automation technology of tomorrow. (Taken from http://www.festo.com/ext/en/12083.htm February 2010)

The penguins’ swimming and diving behaviour has been studied in Antarctica for many years. Using state-of-the-art methods, researchers have succeeded in revealing the secrets of the underwater “flight” of this unusual order of birds. In their search for food, penguins often travel more than a hundred kilometres per day; “Adélie” penguins dive to depths of up to 350 metres, and their larger cousins, the emperor penguins, to as much as 700 metres. In the water they are fast, have a great deal of endurance and are astoundingly agile: they can reach a top speed of almost 30 kilometres per hour, although their speed of travel in the more energy-efficient migratory mode is around 10 to 15 kilometres per hour. Penguins prove robust and crash proof when landing on an iceberg after an audacious leap or making their way through the pack ice. Their artless elegance is matched by the highest levels of energy efficiency and a streamlined body design. Haulage tests with cast models of the spindle-shaped penguins’ bodies show a flow resistance 20 to 30% lower than the hydro - dynamically most favourable known technical bodies (cd-values< 0.02, with Reynolds numbers in the order of 106). The elastically deformable wing surfaces also make for high thrust efficiency.

These phenomenal feats from the animal kingdom provided the inspiration for the bionic realisation of the AquaPenguin.

The bionic penguins are designed as autonomous underwater vehicles (AUVs) that independently orient themselves and navigate through the water basin and develop differentiated, variable behaviour patterns in group operation. The penguins’ hydrodynamic body contours and elegant wing propulsion principle were adopted from their natural archetypes. The wings comprise a skeleton of spring steel elements embedded in an elastic matrix of silicon that gives them their profile; they can thus twist to an optimal angle in interaction with the hydrodynamic forces in each stroke, whereby the pitch angle can also be regulated interactively. The robotic penguins can thus manoeuvre in cramped spatial conditions, turn on the spot when necessary and – unlike their biological archetypes – even swim backwards. An entirely new feature in robotics is the torso that can move in any direction. To make such an “organic” change of shape possible, the head, neck and tail segments were based on a new 3D Fin Ray® structure. This structure, derived from the tail fin of a fish, has thus been extended into three-dimensional space for the first time.

Page 15: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

Figure 12. Rear section as a 3D Fin Ray® structure. (Taken from http://www.festo.com/ext/en/12083.htm February 2010)

The bionic penguins readily demonstrate what is meant by learning from nature. The use of innovative technical materials and the creative combination of various design and functional principles pave the way for new opportunities in design and automation technology. The penguins’ torso design can be used in automation for flexible tripods, thereby opening up new fields of application in handling technology. The BionicTripod has an operating range that by far transcends that of the conventional tripod configuration; for example, pick-and-place applications with an offset of 90 degrees are possible. In combination with a flexible and adaptive gripper, fragile objects of various shapes can be moved. With the AquaPenguin, Festo is benefiting from the advantages of versatile contour and structural adaptation and intelligent self organisation, both on an individual level and in group operation.

Page 16: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

References

[1] Barras, C., 2001, Learning to love to hate robots. New Scientist, 204, 2738, p.22.

[2] Anon. 2007, Study: Roomba Owners Get Deeply Attached to Robots. Fox News [online], Available at: http://www.foxnews.com/story/0,2933,299022,00.html. Last Accessed: 14th February, 2010.

[3] Anon. 2006, Pennsylvania Hospitals Purchase Aethon's Tug. Go Robotics [online], Available at: http://www.gorobotics.net/the-news/latest-news/pennsylvania-hospitals-purchase-aethons-tug/. Last Accessed: 14th February, 2010.

[4] Anon. 2008, Toward a Smarter Bot. Mechanical Engineering, 130, 7, p.17.

[5] Anon. Brown, A.S., (ed). 2008, What Robots Learn. Mechanical Engineering, 130, 11, p.22.

[6] Mitri, S., Floreaon, D., \& Keller, L., 2009, The evolution of information supressing in communicating robots with conflicting interests. [online], Avaliable at: http://infoscience.epfl.ch/record/139388/files/. Last Accessed: 15th February, 2010.

[7] Chew, C.M., Pratt, J., \& Pratt, G., 1999, Blind Walking of a Planar Bipedal Robot on Sloped Terrain. Proceedings of the Internation Conference on Robotics and Automation, Detroit, Michigan.

[8] Hobbelen, D., De Boer, T., \& Wisse, M., 2008, System overview of bipedal robots Flame and TUlip: tailor-made for Limit Cycle Walking. IEEE/RJS International Conference of Intelligent Robotics and Systems, Nice, France.

[9] Meijer, R., 2008, Walking Like A Human: TU Delft Robot Flame. Medical News Today, [online], Available at: http://www.medicalnewstoday.com/articles/108605.php Last Accessed: 20th February, 2010.

[10] Vaidyanathan, V., \& Sivaramakrishnan, R., 2007, Department of Production Technology, Madras Institute of Technology, Anna University, INDIA. Design, Fabrication and Analysis of Bipedal Walking Robot. [online], Available at: http://www.robogames.net/papers/2007/07-108-Vaidyanathan-AnnaUniv-AnalysisofBipedalWalkingRobot.pdf Last Accessed: 20th February, 2010.

[11] Whitby, B., 1997. Why The Turing Test is AI's Biggest Blind Alley. [online] Available at: http://www.informatics.sussex.ac.uk/users/blayw/tt.html Last accessed: 26th February

[12] Biologically Inspired Robots, Fred Delcomyn

[13] http://www.universalleonardo.org/, Univeral Leonardo website.

[14] Viana, F.A.C., Steffen, V. Jr., Butkewitsch, S., & de Freitas Leal, M., 2009. Optimization of aircraft structural components by using nature-inspired algorithms and multi-fidelity approximations. Journal of Global Optimization, (45), p.427-449.

[15] Self-Organization, Embodiment, and Biologically Inspired Robotics, Rolf Pfeifer, Max Lungarella and Fumiya Iidal.

[16] http://web.mit.edu/newsoffice/2009/stickybot-092509.html, MIT News, September 2009.

[17] http://homepages.inf.ed.ac.uk/bwebb/, Barbara Webb’s Homepage.

Page 17: Nature Inspired Robotics - University of Birminghamrjh/courses/NatureInspiredDesign/2009-10... · robotic vacuum cleaner, and TUG, a medical supplies transporter. Roomba is a small,

[18] Robot Phonotaxis in the Wild: a Biologically Inspired Approach to Outdoor Sound Localisation. A.D. Horchler, R.E. Reeve, B.H. Webb, R.D. Quinn. 2004.

[19] New neural circuits for robot phonotaxis. Richard E. Reeve and Barbara H. Webb. 2003.

[20] Robots, crickets and ants: models of neural control of chemotaxis and phonotaxis. Barbara Webb. 1998.

[21] Rationale behind Biological Inspiration in Robot Design, Satyandra K. Gupta

[22] “Hexapodal Robot Locomotion Over Uneven Terrain”, Barnes, D. P, Proc. IEEE Conf. on Control Applications. Trieste, Italy, pp. 441 – 445, September 1998.

[23]“Artificial Neural Nets for Controlling a 6-legged Walking System”, Cruse, H., Müller-Wilm, U., Dean, J, Proc. of the Second International Conference on Simulation of Adaptive Behaviour: From Animals to Animats 2, MIT Press, Cambridge, MA, pp. 52 – 60, 1993.

[24] “What Mechanisms Coordinate Leg Movement in Walking Arthropods?”, Cruse, H, Trends in Neurosciences, Vol. 13, pp. 15 – 21, 1990.

[25]. Festo AG & Co. KG Corporate Design Ostfildern Germany www.festo.com/bionic, February 2010.