Walking – following in nature's footsteps

Citation

Virk, G.S. (2002), "Walking – following in nature's footsteps", Industrial Robot, Vol. 29 No. 3. https://doi.org/10.1108/ir.2002.04929caa.002

Walking – following in nature's footsteps

Walking – following in nature's footsteps

Gurvinder S. Virk

Walking is an extremely difficult operation and it has not yet been possible to design and build a really good walking robot. However, walking is a fundamental capability in nature and most land-based creatures have "come to the conclusion" that walking is the best form of locomotion. Robotics generally, and mobile robotics in particular, has taken much inspiration from biology, but it seems clear that the robot design community has failed dismally in comparison to the major successes seen in the natural world. The robust and reliable solutions evolved over centuries are light years ahead of the robots produced to date. It is clear that much work is still needed to first of all understand the problems, investigate and determine how the biological solutions work, and determine if these strategies can be adopted and implemented in robotic solutions.

In this issue, the theme is "Exploration and inspection" and what is the relevance of walking robotic systems. To form an opinion of this, we should look at nature, as all creatures must explore their environments in searching for food and for mates. "Inspection" is also carried out because the creatures do not venture into areas that are dangerous for them. In view of this and the above stated fact that nearly all land-based creatures walk, we must come to the inevitable conclusion that walking robots must be optimal. However, robotic technology is not yet ready to deliver the solutions that can perform at the required level and so we must be content with sub-optimal solutions.

The situation is compounded by the fact that robots do not need to be general-purpose machines and are normally designed for specific tasks to be carried out in certain environments. For example, exploration and inspection tasks that can be carried out by robotic systems include the following:

• •

  Disaster scenarios and searching for victims.

• •

  Inspecting and maintaining buildings and other structures, industrial plant and systems (in air, underwater and in the presence of various chemicals).

• •

  Humanitarian land mine clearing.

• •

  Medical systems to explore and treat patients.

• •

  Mining and underground applications.

• •

  Outdoor/natural environment applications.

To carry out these tasks in a reliable manner requires sophisticated robots, and these simply do not exist. Wheeled or tracked robots are currently the most reliable and hence the reason for their use to carry out some of the above tasks to a certain level of satisfaction. Such machines allow efficient movement in 2D environments normally encountered in indoor situations. It is argued that the robots can even move around in rough outdoor terrains, although the going can get quite shaky when the wheeled machines try to move fast. Under such cases it is necessary to go for more sophisticated systems that have legs as "discovered" by nature. Even in indoor applications wheeled robots can be ineffective if the floor is very cluttered with obstacles. In some cases robots need to climb structures to perform the desired exploration and inspection operations. To do this, machines need an appropriate climbing capability. This can be achieved by using vacuum suction (for non-porous surfaces), magnetic adhesion (for ferrous surfaces), or have specialised end effectors (for gripping on to protrusions or crevices in the climbing surface being explored or inspected).

As in other areas, the limitation of robotics has meant that compromises have to be made. Sometimes this is referred to as the "80 per cent rule". This is where the solutions are designed to achieve only 80 per cent of the total requirements because this level is normally achievable quite easily, and the remaining 20 per cent is usually very difficult and not worth bothering with because of the poor cost/benefit analysis.

It is probable that most people would see walking robots as lying in the "difficult 20 per cent". Whether this is right or not is clearly a debating issue, but economics will always win and over-engineered systems will not be encouraged in the foreseeable future. Maybe it is necessary for the technology to take a major step forward in all the areas that are needed to achieve solutions that can rival those in the biological world. When this happens, then perhaps the 80/20 rule will be ousted and customers will look for machines that will last and survive all conceivable situations rather than merely do the majority of the work. Clearly, much work is needed to get over this blinkered approach.

The 80 per cent rule is detrimental from another viewpoint because it does not seek to stretch the technologies where they are deficient. It simply puts all the difficult aspects of the requirement into the unachievable area and does not even try to address them. Consequently, systems that are produced are unimpressive, fail to inspire and demand for them is always relatively low.

In addressing issues of walking robots for exploration and inspection, it is clear to me that we are looking at the 20 per cent areas, and as such, we are trying to push back the technology frontiers so that major breakthroughs can be made. Other users are beginning to become interested in walkers because the "80 per cent machine market" is becoming saturated and so robot developers are attempting to find ways of obtaining a competitive edge.

In fact machine locomotion does not need to just copy nature in the clear-cut manner that I have tried to make out here. We do not just have wheeled systems or walking systems as the only options available. There is a whole range of locomotion capabilities that can be looked at and the most appropriate level selected. In doing some analysis of this for space exploration, the consortium on climbing and walking robots (CLAWAR) within Europe (this is an EC project that I am currently co-ordinating, see URL www.uwe.ac.uk/clawar) has broken down locomotion into nine levels (the simplest at level one and the most complex at level nine). Although by no means unique, it is an interesting way of looking at mobile robot locomotion. The nine levels are as follows:

1. 1.

   Wheeled and tracked machines for moving on flattish or moderately uneven terrain.

2. 2.

   Passive bogies for simply control and ability to negotiate some roughness in an uneven terrain.

3. 3.

   Active bogies for more challenging terrain.

(These three locomotion methods require continuous paths for contact with the surface being negotiated. Legged systems, on the other hand, require discrete points of contact. This of course can be a very useful capability in many environments.)

1. 1.

   Simple legged robots with limited degrees of freedom.

2. 2.

   Legged/wheeled hybrid systems. Here, wheels are used for flattish environments and the legs are used to step over obstacles.

3. 3.

   Climbing robots for artificial surfaces in terrestrial environments, such as buildings.

4. 4.

   Articulated multi-jointed robots with multiple degrees of freedom.

5. 5.

   Climbing robots for natural terrains.

6. 6.

   Universal walking and climbing machines for natural terrain.

I believe that the robot technology is roughly at level seven), although the machines that have been produced are of a largely prototype nature and considerable work is needed still for the machines to be further developed and commercially exploited as viable and reliable alternatives to the existing (non-robotic) solutions that are currently being adopted.

Robots for exploration and for inspection do need a walking capability in general, and in many specific applications they also need a climbing capability. Walking is an important capability that needs further work. The "20 per cent difficult area" is clearly becoming the natural focus for R&D and as such, walkers must be a key area to focus on. Finding ways of producing walking machines for different types of surfaces is needed urgently. In walking, many aspects need to be considered. How many legs should the machine have? How many joints must each leg have? Should all the legs be identical or should they vary depending on the position in the body? What type of walking gait should be adopted? How should the gait be generated? Can this be adaptive and change automatically to changing environments? Again, it is possible to look at nature and get some insights to see that there is normally symmetry in the axes of motion. Here, modularity could also be important in that the left and the right side of the machine should be symmetric.

In conclusion, it seems wishful to think that walking machines can play a useful role in exploration and inspection, but this is unlikely to happen in the foreseeable future. The first thing that must happen is that the reliability of walking robots must improve, possibly by looking at applications that are not safety critical. This is already happening by the concentration on the edutainment sector with all the walking toys appearing on the market. Hopefully, this will have the required effect and we will soon have a mature walking robot technology that can be used in more serious and potentially beneficial applications such as those described here.