Icon for: Dwight Springthorpe

DWIGHT SPRINGTHORPE

University of California at Berkeley
Years in Grad School: 2
Judges’
Choice

Judges’ Queries and Presenter’s Replies

  • May 21, 2013 | 09:22 a.m.

    Fascinating.
    given what you have already learned, can you think of a specific new technology that could be developed based on your findings?

  • May 21, 2013 | 07:13 p.m.

    I believe there are several potential applications, particularly within the field of robotics. For instance, we are currently working with our collaborators (Prof. Koditscheck’s lab at Univ. of Pennsylvania) to implement the basic burrowing mechanism (hook-and-pull) into an existing hexapedal robotics platform (a modified version of the RHEX robot- see reference below). Specifically, we believe the c-shaped leg, common to many running robots, can be modified to include an additional joint around the middle of the leg. The articulation should enable the robot to recreate the basic hook-and-pull motion, creating a robot that can run and manipulate the terrain.

    .

    After we have developed the basic burrowing mechanism, implementing the crabs’ burrowing posture and ability to rotate within the burrow could also substantially increase the robot’s ability to manipulate the environment by giving the legs a much larger working space. We believe that the result of all this work will create a robot that has an unprecedented ability to move above and within complex terrain, such as collapsed buildings.

    .

    More broadly, we also believe that studying animal multifunctionality can make conceptual contributions to future technologies. Our preliminary work suggests there are many components that enable multifunctionality, including flexible control strategies, highly articulated limbs and behavioral adaptations (i.e. learning). As we continue to study this system, we hope to learn exactly what each of these features contribute to the capabilities of the ghost crabs and allow us to understand what features are most critical to multifunctionality. Overall, we hope to contribute broadly to robotics and any other technology that can benefit from multiple operating modes.

    .

    Reference: X-RHex: A Highly Mobile Hexapedal Robot for Sensorimotor Tasks. Kevin C. Galloway, G. C. Haynes, B. Deniz Ilhan, Aaron M. Johnson, Ryan Knopf, Goran Lynch, Benjamin Plotnick, Mackenzie White, D. E. Koditschek University of Pennsylvania Technical Report, 2010

  • May 21, 2013 | 11:29 a.m.

    What sensory cues does the crab use to guide these movements, and to what degree do they seem to be modifiable on a moment by moment basis from available cues?

  • May 21, 2013 | 07:43 p.m.

    At present, we cannot say conclusively which sensory cues are most important to the burrowing behavior. Anatomical references (e.g. G.F. Warner, The Biology of Crabs) indicate that the crabs have a diverse sensory suite that includes the ability to transduce touch, balance, limb position and limb force – all of which could be of importance to burrowing. By using wireless telemetry systems (see reference below), we believe we can access many of these signals and correlate them with environmental manipulations (e.g. moisture content), allowing us to determine which sensory cues result in behavioral variation.
    .
    Some of our current work suggests that there is some sensory feedback embedded within this behavior. For instance, we have observed crabs burrowing against the wall of the container. Here, the crabs try several times to collect material from the container wall, failing to collect any material. Then, the crabs perform a body rotation followed by more excavation. Although more work is required, this does suggest that the crabs can modify the burrowing behavior on a moment-to-moment basis.
    .
    Reference: “A miniature animal-computer interface for use with free-flying moths.” Dwight Springthorpe. Master’s thesis, University of North Carolina, Chapel Hill. 2009.

  • May 21, 2013 | 06:55 p.m.

    In your poster you state that burrowing behavior does not vary significantly among the individuals tested. I am curious about where the variation occurs, can you please describe the variation that you observed and comment on whether you think the behavior is learned or genetically encoded.

  • May 22, 2013 | 12:23 a.m.

    For the individuals and metrics used here (hook-and-pull duration, hook-and-pulls per excavation event and excavation rate), we did not find any statistically significant variation between individuals engaged in vertical burrowing. However, we did observe some significant differences when we compared horizontal and vertical burrowing. Although the hook-and-pull duration and the number of hook-and-pulls per event remained the same, the crabs were able to excavate significantly faster. Overall, this difference seemed to be predominantly due to reduced resting durations, which may be because vertical burrowing is more energy intensive.
    .
    While more work is required, especially regarding how the kinematics and burrowing forces change with material properties, this does suggest that the crabs can modify the burrowing behavior to accommodate different conditions. Additionally, the observed lack of variation may be because the crabs were all nearly the same size and burrowing in equivalent materials, potentially indicating that the observed parameters are tuned to the studied sizes and materials. We intend to follow up on this idea with future experiments where we examine how burrowing kinematics, forces and muscle activity change in response to different conditions, such as water content and packing density. Given that small changes in material properties can produce large changes in interaction forces (see reference #1), we expect to observe some changes. However, it is possible that the burrowing behavior is mechanically robust and will not vary in response to some perturbations, which has been observed in other invertebrates (see reference #2).
    .
    You also raise a very good point about learned vs. innate behaviors. Given the life history of these crabs (no parental contact, limited social interaction), we believe that burrowing probably builds on an innate behavior. However, the crabs can burrow in a variety of substrates and further work may reveal that there is a learned component, similar to learned changes in innate behaviors demonstrated in other invertebrates (see reference #3). As we proceed, we intend to carefully track individual changes in behavior over time to see how learning may influence burrowing. These results may be particularly interesting since learning new tasks may be a major component of multifunctionality.
    .
    Reference #1: Gravish, N., Umbanhowar, P. B., & Goldman, D. I. Force and Flow Transition in Plowed Granular Media. Physical Review Letters, 105(12), 128301. 2010.
    Reference #2: Sponberg, S, Full RJ. “Neuromechanical response of musculo-skeletal structures in cockroaches during rapid running on rough terrain.” J Exp Biol. 211: 433-46. 2008.
    Reference #3: Menzel,R., B.Brembs, and M.Giurfa. “Cognition in Invertebrates.” Evolution of Nervous Systems, Vol. II: Evolution of Nervous Systems in Invertebrates (ed. J.H.Kaas), Chapter No. 1.26, pp. 403-422. Academic Press, Oxford. 2007.

  • Icon for: Mary Gauvain

    Mary Gauvain

    Judge
    May 21, 2013 | 10:39 p.m.

    These results suggest that multimodal tools are useful in adapting to different situations. Extrapolate these results in a way that provides insight into the use of these findings for developing technologies will solve currently unsolved problems?

  • May 22, 2013 | 12:59 p.m.

    You make an excellent point. We agree that multimodal appendages are important for animal multifunctionality. Thus, we ask: what features permit the greatest degree of multifunctionality? Our observations already suggest some features which we intend to explore further.
    .
    For instance, it seems that burrowing crabs use the same muscles used during running. However, the leg movements we observed suggest that burrowing crabs employ a very different pattern of activation than running crabs. Future work, examining muscle activation and force production, will further explore the differences between the two functional modes. Although more work is required, the observations presented here indicate that burrowing crabs use their legs to slowly develop comparatively large forces. Running crabs, however, either move their legs quickly with minimal resistance (during the aerial phase of running) or use their muscles as variable-stiffness springs (see reference #1). This suggests that flexibility in muscle ‘gearing’ is a major component of multifunctionality.
    .
    We also observed that the crabs’ legs have many degrees of freedom (DOF), which are used differently during running and burrowing. Running crabs use a small fraction of the total workspace, moving mostly within a plane (see reference #2). On the other hand, burrowing crabs appear to use the DOF differently and access much larger portions of the workspace, even using body rotation to expand the workspace to areas that are unreachable during running. This seems to indicate that highly articulated limbs enable multimodal operation. In the future, we intend to explore this idea in more depth by comparing the relative importance of different DOF during running, burrowing, jumping and manipulation. We expect to find that some articulations and workspace dimensions are particularly valuable to multifunctional behavior.
    .
    These findings will translate into new technologies that can address unsolved problems in robotics. For example, we intend to add motor flexibility and additional articulations to existing bio-inspired running robots, such as RHEX, that have already implemented spring-loaded running with only single-jointed legs (see reference #3). This may substantially improve the robot’s ability, creating a single multi-purpose platform capable of diverse tasks, such as jumping, moving fluidly across irregular terrain and environmental manipulation (e.g. burrowing).
    .
    Reference #1: “Exoskeletal strain: evidence for a trot-gallop transition in rapid running ghost crabs.” Blickhan, R., Full, R.J. and Ting, L.H. J. exp Bio. 179, 301-321. 1993
    Reference #2: “Underwater punting by an intertidal crab: a novel gait revealed by the kinematics of pedestrian locomotion in air versus water.” Martinez, M., Full, R., & Koehl, M. The Journal of experimental biology, 201 (Pt 18, 2609–23. (1998)
    Reference #3: “X-RHex: A Highly Mobile Hexapedal Robot for Sensorimotor Tasks.” Kevin C. Galloway, G. C. Haynes, B. Deniz Ilhan, Aaron M. Johnson, Ryan Knopf, Goran Lynch, Benjamin Plotnick, Mackenzie White, D. E. Koditschek University of Pennsylvania Technical Report, 2010

  • May 21, 2013 | 10:42 p.m.

    Interesting. I love the simple perfection and efficiency of complex biological systems. Multi-funcitonality is a great example of that simplicity in the switching of tasks. Why did you not measure task switching directly? What data did you collect that will improve biomimicry and multi-funcitonality in robotic applications?

  • May 22, 2013 | 02:17 p.m.

    We agree that task switching is an interesting and important component of animal multifunctionality. After all, an animal cannot be multifunctional if it cannot effectively switch between different modes of operation. We certainly intend to follow up on this idea, particularly relating excavation behaviors (e.g. switching between hook-and-pull and scratch digging) to the properties of the environment (e.g. substrate properties and obstacles). However, before we could work on that, we required a basic understanding of the overall behavior, which we present here. As we continue, we will further quantify the behavior with wireless telemetry devices that will give us more detailed access to kinematics, forces and muscle activation (see reference #1). This will allow us to examine the ‘how’ and ‘why’ of mode switching. Additionally, this expanded data set may reveal modes that are kinematically similar (and thus, indistinguishable with videographic methods) but with very different force production, as was found in running and climbing cockroaches (see reference #2). By combining the detailed behavioral description we present here with these future experiments, we believe we can more completely address the topic of task switching.
    .
    Much of the data we collected here can translate into improved bio-inspired robots. For instance, the crabs demonstrate several mechanisms showing how a leg ‘designed’ for running can also be used to manipulate the environment. We are currently working with our collaborators (Prof. Koditscheck’s lab at Univ. of Pennsylvania) to recreate the essential features of these mechanisms in a bio-inspired running robot, leading to a multifunctional platform capable of both rapid running and environmental manipulation (see my response to Ayelet Gneezy’s question for additional detail). As we continue to examine this behavior in more depth, we believe that other aspects of burrowing will inspire further improvements. Task switching, for instance, may offer a model for a high-level algorithms that mediate transitions between a robot’s different functional modes. Overall, burrowing represents an interaction between biomechanics, behavior and the environment. By understanding how these features interact and which features are most important to multifunctionality, we hope to contribute to robotics technology both specifically, by inspiring designs or movement patterns, and conceptually, by building a theoretical framework of multifunctionality.
    .
    Reference #1: “A miniature animal-computer interface for use with free-flying moths.” Dwight Springthorpe. Master’s thesis, University of North Carolina, Chapel Hill. 2009.
    Reference #2: “Dynamics of rapid vertical climbing in a cockroach reveals a template.” Goldman, D; Chen, T; Dudek, D; Full R. The Journal of Experimental Biology.” 209, 2990-3000. 2006.

  • May 22, 2013 | 09:47 p.m.

    Thank you Dwight!

  • Further posting is closed as the competition has ended.

Presentation Discussion
  • May 21, 2013 | 10:43 a.m.

    I enjoyed the video! The X-ray videos are so cool!!

  • May 21, 2013 | 07:16 p.m.

    Thanks! I was impressed by your work as well.

  • May 21, 2013 | 11:35 p.m.

    This may be a silly question, but does the x-ray technique damage the crabs in any fashion? Is it low level radiation, or do the crabs have a resistance to this type of exposure. This is a really cool technique!

  • May 22, 2013 | 12:36 a.m.

    Not a silly question at all. Animal welfare is important and, if the x-ray system was severely detrimental to the crabs, then we might obtain aberrant results.

    The x-ray machine we used was a fairly standard fluoroscope, similar to what a doctor would use to examine an injured limb. Our videos are similar in length to the exposure associated with some fluoroscope-aided surgeries. This, coupled with the shielding that the sand provides, leads us to believe the crabs did not suffer any significant detrimental effects.

  • Further posting is closed as the competition has ended.