Robustness in legged systems

As part of the graduate programme at the RVC (Royal Veterinary College), we give talks every year of our PhD in the Internal Seminar series. I was due to give mine today – but as I sit here snotting and coughing over my keyboard, I can’t help but feel that the people of the RVC will be glad that I had to cancel.

Yes, I have caught what seems like yet ANOTHER horrid cold. Snotty, coughy, feverish – the full works. And most upsettingly, I’ve lost my voice. So no seminar talk for me today. But instead, I thought I’d write about all the things I was going to say! It’s pretty much just an overview of my PhD so far and its aims, and as I keep promising a blog post on such things, it seemed like a good opportunity to do so!

So, first: the full title of my PhD – “Robustness in legged systems: an integrative approach.” It’s a nice title, because it pretty much left it up to me what ‘legged system’ I wanted to explore. As most of you will know, I chose the very-many-legged spider as my study animal. The “integrative approach” bit means that I bring together stuff from different disciplines to talk about robustness – in my case, biomechanics (spiders) and robotics!

Robustness is a tricky thing to define, both in layman’s terms and technically. Applying it to what I do, I tend to think of robustness as a measure of resistance to change, or an ability to cope with or adapt to change. For example, spiders have pretty robust locomotion (as do all legged animals) – if obstacles or bumpy ground are introduced to their environment, they have no problems running over it. Equally, if they lose a leg, they (generally) have no problems adapting their gait. And when talking about robustness in legged robots, these are the kinds of things I think about, too.

Why spiders? Apart from the fact that they’re awesome, you mean?! But really, there were several reasons for studying spiders…

  1. As a species, they live across all terrain types, so they’re highly adaptable.
  2. They can naturally lose legs in a sort of self-amputative process called autotomy, meaning that I can study how their gait adapts to missing legs. We know it must adapt somehow, because although some spiders can grow the legs back (depending on their age & species), sometimes they don’t grow back – so the spider must be able to move around for the rest of its life with less than eight legs. I’m super curious as to how they can do this, especially immediately after the leg has been lost.
  3. Spiders are pretty easy to get hold of! In the UK, wolf spiders are easy to spot in grassy areas in May/June, and garden spiders are abundant at the end of the summer in their massive webs. Always make your PhD life a bit easier if you can!!

Spider gait: Spiders move their legs in alternating tetrapods. This is quite difficult to explain in words, so here’s a picture to help me:


The blue legs form one tetrapod, and the red legs form the second tetrapod. When a spider moves, its most basic gait is simply moving all the red legs together, and then all the blue legs together. See how they’re diagonally criss-crossed over the body so the centre of mass is constantly supported by the legs. This pattern was well described and discussed by Donald M. Wilson in his 1967 paper about tarantulas. In this paper, he also looked at the effects of losing two legs from the same tetrapod – how does the spider adapt its gait? Of course, in 1967 technology to quantify this was limited, and so the paper is more descriptive than data-heavy.

So the aims for my PhD, with the age of high-speed cameras and automated tracking software, were to look at this in more detail – how exactly do spiders change their gait when they lose legs from the same tetrapod? How do their legs rearrange (if at all)? How long does it take them to adapt – do they switch to their ‘new’ gait straight away, or is there an intermediate gait first? And, finally, how can we use this knowledge to improve our legged robotics?

What have I done so far? I’ve collected a large dataset from the common garden spider, Araneus diadematus. These spiders predominantly move around on webs but are also able to move over ground. I looked at how they adapt their gait to missing legs by inducing autotomy in two legs within one tetrapod. High-speed footage taken from above as the spider runs across a flat perspex arena allows me to track individual legs, producing XY coordinates. These can be used to work out what gait the spider is using.

Digitising the high-speed videos like this is proving to be an arduous task – spiders have a lot of legs! So I’m still in the process of doing this. But preliminary results from two of the spiders look like they might be doing some funky things with their front & hind legs… Watch this space!

What am I doing now? Currently I’m working on collecting a dataset from wolf spiders – repeating the same experiments as with the garden spiders, but with added Lego rough terrain! (See my previous post if you’re interested!) This is so I can directly compare the two species. I added the rough terrain with wolf spiders because they’re hunters, actively hunting their prey down on the ground, so they should be well adapted to moving over rough terrain, and I wanted to see how they manage it after autotomy. As well as this, I’m still digitising the garden spider work, and preparing for a poster presentation at the Society for Experimental Biology’s annual conference this summer.

The final steps of my PhD are to learn from the spiders and apply it to robotics! I’m currently awaiting delivery of a T8X spider robot, which I can’t wait to get my hands on! This particular robot is marketed as a robot for everyone – it’s ‘plug and play’ so amateur roboticists can play straight from the box, but the programming is also accessible so you can hack into it to change different things – such as leg timings. This is what I’m hoping to be able to do – to design a new spider controller based on the data collected in actual spiders, that allows the robot to adapt its gait if legs are missing or damaged – to become robust to leg loss. This has applications in robots that are used in search & rescue situations, in the armed forces, and even perhaps in space exploration. Anything that lets the robot carry on its merry way without having to be taken in to a workshop to be fixed is a bonus!

So, that’s it! That’s where I’m at so far, and what’s left to do (eek!). Apologies to the people at the RVC who organise the seminar series – I’m not sure yet (at time of writing) whether my talk will be reallocated to another date, but thanks for understanding.

I genuinely am really looking forward to writing this up in my thesis!

Thanks to BBSRC for funding me, and to my supervisors for putting up with and getting me through my numerous blips so far!

If you enjoyed reading this, you might also like: Spiders and Lego


3 thoughts on “Robustness in legged systems

  1. Pingback: SEB Conference: A review | There's A Spider In The Bath!

  2. Pingback: Pint of Science… With Robots | There's A Spider In The Bath!

  3. Pingback: Pint of Spiders… With Robots | There's A Spider In The Bath!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s