Tailbot – Robot with Inertial Assisted Control

View Video View Poster

See Related: Biology Engineering Sensing & Signals

EVAN CHANG-SIU

Presenter’s IGERT

University of California at Berkeley

Years in Grad School: 5

Judges’
Choice Community
Choice

Tailbot – Robot with Inertial Assisted Control by an Active Tail Inspired by Lizards

Lizards, discovered to pitch correct in mid-air with their tail when subjected to slippery take-off surfaces, have inspired a novel approach to stabilizing rapid locomotion in mobile terrestrial robots. To demonstrate the benefit of this behavior we built a 177g wheeled robot, Tailbot, with inertial sensors, a microprocessor, and a single degree-of-freedom active tail. Since the tails effective inertia scaled quadratically with length, its mass was designed to be less than 20% of the body mass while still allowing for a one to one ratio of angular stroke. By estimating the body angle from the inertial sensors and utilizing both contact forces and zero net angular momentum maneuvering, Tailbot could take advantage of closed loop feedback control. Feedback produced rapid reorientation during a fall, smooth transitions between surfaces of differing slopes, and stability when faced with perturbations that would overturn a tailless robot. Specifically, Tailbot could perform a 90o self-righting maneuver during free fall in 138ms corresponding to a drop distance of approximately one body length. A nominally catastrophic perturbation, produced a 60o rotation in a passive tailed robot, but resulted in only a 30o rotation in our feedback controlled tailed robot. Landing transitions that were not possible with a tailless robot were made feasible by properly adjusting the reference angle to the tail controller. Capabilities of Tailbot demonstrate how active tails can improve the stability and maneuverability of terrestrial and aerial search-and-rescue vehicles and serve as a physical model to generate new hypotheses of inertial appendage control in animals.

Abstract
Share

Tailbot – Robot with Inertial Assisted Control by an Active Tail Inspired by Lizards

13414 Views

Share Presentation

Link

Judges’ Queries (11) Discussion (12)

Judges’ Queries and Presenter’s Replies

Judges and Presenters may log in to read queries and replies.

Presentation Discussion

Kathryn Hobbs

Project Staff

May 22, 2012 | 03:32 p.m.

Fascinating work! I enjoyed watching your video – your explanation together with the included footage made it very clear.
Thank you!
Evan Chang-Siu

Lead Presenter

May 23, 2012 | 12:42 a.m.

Thanks!
Haifei Cheng

Guest

May 23, 2012 | 12:51 a.m.

Nice work!
garry siu

Guest

May 23, 2012 | 10:18 a.m.

A very informative video, hopefully the active inertial sensor will lead to further developments and incorporated into terrestrial and aerial search and rescue vehicles.
Debra Bernstein

Project Associate

May 23, 2012 | 03:12 p.m.

Great video! (I particularly love the shot of the lizard sitting on your robot). I’m curious about the practical applications of your work. In the video you mentioned search and rescue robots – I can see the tail being useful in those situations, although it also seems like an appendage might just get caught in debris (not sure if that’s a concern or not). Are there other ways you could see applying this type of design?
Evan Chang-Siu

Lead Presenter

May 23, 2012 | 03:47 p.m.

Thanks! Your insight is correct, and I totally agree that the cons of having a tail could be that it gets caught on debris. This would definitely be one reason to limit the length of the tail even though a longer tail would be more effective in terms of providing a great moment of inertia. However, biological systems such as lizards, cats, lemurs do have tail lengths that are up to and greater than body length and fair very well in unstructured environments. What this may call for in an engineering setting is a flexible tail design as opposed to a purely rigid one that can deform around obstacles or maybe one that detaches easily as a last resort in emergency situations. Depending on the task of the robot, it would be nice to be able to quantify how long is too long for a tail, but I think this is a very complicated problem. Other applications would definitely fall into air or space vehicles, since orientation is very important.
Debra Bernstein

Project Associate

May 25, 2012 | 09:51 a.m.

Thanks, Evan, for your reply. It’s really interesting to see how you’re mapping between the biological and engineered worlds.
Amber Bratcher

Graduate Student

Amber's video »

May 23, 2012 | 04:40 p.m.

Nice work! I am also curious about the practical applications of having a tail on a robot. In terms of encountering obstacles on the ground, couldn’t instability issues be overcome by modifying the wheel or track pattern used for locomotion? I’m wondering about the costs of having a tail in terms of power requirements and additional weight, as well as the monetary costs of the additional sensors needed to manipulate the tail. Or perhaps these costs are so minimal that they are not an issue. I can definitely see the advantage to a tail when flying through the air, but I don’t know how often flying through mid air would occur for a rescue robot. Regardless, it’s really interesting work!
Evan Chang-Siu

Lead Presenter

May 24, 2012 | 12:39 a.m.

Thanks! I agree that modifying the mechanisms that affect ground interaction and designing intelligent strategies will definitely help with stability. However, we feel that the next generation of search and rescue robots will be designed to quickly run over unstructured terrain unlike the slow moving ones of today. You can imagine this fast running would cause many perturbations, where an active tail could quickly whip around to perform an inertial maneuver or even reach out and contact the ground to compensate.

Experiment 3 at time 1:13 in this video shows a simple case:
http://www.youtube.com/watch?v=s2Lk_2YCtA4

We are very interested in (and currently exploring) how the ground interaction forces and the inertial forces can be coordinated so that in certain cases we can get improved stability (disturbance rejection) or in other cases improved maneuverability (fast turning).

Added weight is definitely a factor, and we have shown that we can make an effective tail with motors and transmission to be less than 20% of the body, which is about the same ratio as the lizards tail to body mass ratio:

http://ieeexplore.ieee.org/xpl/articleDetails.j...

As for cost, inertial measurement units are becoming much cheaper and smaller these days. For example you can buy a Nintendo wii mote, which has all the necessary sensors for under 30$.

Thanks for your thoughtful questions!
Carol Johnson

Graduate Student

May 24, 2012 | 01:46 p.m.

Neat stuff! Sounds like it’s a lot of fun. Great presentation, very clear and easy to understand.
Evan Chang-Siu

Lead Presenter

May 24, 2012 | 02:57 p.m.

Thanks!
Christie Peters

Guest

May 25, 2012 | 01:39 p.m.

Any project that can contribute to search and rescue vehicle design AND provide insight into the biomechanics of velociraptors gets my vote!
Further posting is closed as the event has ended.