PDA

View Full Version : Re: Asimo robot gait



d3uk61
12-07-2007, 10:31 PM
Stuart wrote

Quote

"A straight knee is a singularity that reduces the rank of the
kinematic jacobian, meaning that the foot velocity cannot have a component
parallel to the line connecting the hip and ankle when the knee is
straight. ----------------------some
cut-------------------------------------

An alternative approach that uses straight knees during stance and seems
more closely tied to biomechanics is passive dynamic walking as demonstrated
by McGeer, Ruina, Kuo, and Wisse. This article in science news provided a
good overview of the field:"



Stuart

I don't pretend to understand the math of the "jacobian" could you explain
the concept a little more.

Would you say that the straight knee approach is another example of
simplifying the human gait model. This model uses the momentum of the CoM
Plus a small added energy input, in the robot case from electric motors,
which is much more energy efficient than the Asimo model.

However the straight leg Rocker foot model may have far less stability and I
doubt it can climb stairs of cover uneven or rough terrain.

As a person runs over flat ground they increase their efficiency of forward
motion and by using more of the elastic energy stored in the Achilles
tendon. In a similar way Macropods / Marsupials or Kangaroos in particular
have a more efficient forward motion when they bounce than when they walk.

It would appear that Asimo becomes less efficient at higher gait velocities
but increases stability. The rocker gait robot on the other hand becomes
potentially less stable at in creased gait velocities IE if while rocking
over it's stance foot it clipped the contralateral leg during swing thru the
robot would not recover in time to stop it falling on its face. This would
sem quite likely on rough terrain.

The human can of course combine the efficient straight leg rocker action
with the bent knee stable posture and vary the mix according to enviroment,
anatomy, physiology / pathology, and performance.

Do you think a combination of the Asimo and the rocker foot robot would be a
possibility that would be more analogous to 'normal or usual' human gait. EG
it could recognise rough terrain or obstacles and converts from the
efficient stiff knee rocker foot gait to the bent knee stable posture
progression at the appropriate time.

Just my thoughts

Cheers Dave Smith

----- Original Message -----
From: "Stuart O Anderson"
To:
Sent: Friday, December 07, 2007 7:47 PM
Subject: Re: [BIOMCH-L] Asimo robot gait


> Hello,
>
> By way of introduction, I am a Ph.D. student at the Carnegie Mellon
> Robotics
> Institute, where I work with humanoid robots performing gait and balance
> tasks.
>
> It is generally understood that the bent knees exhibited by humanoid
> robots
> like ASIMO or HRP-2 are an artifact of the algorithms used to generate a
> stable gait. A straight knee is a singularity that reduces the rank of
> the
> kinematic jacobian, meaning that the foot velocity cannot have a component
> parallel to the line connecting the hip and ankle when the knee is
> straight. These robots control their balance during gait by accelerating
> their center of mass using the inverse jacobian to determine necessary
> joint
> accelerations. When the jacobian is near a singularity the inverse can
> become badly conditioned, resulting in large joint accelerations or no
> solutions. To avoid this, the robots keep their knees bent during gait.
> This approach to generating stable gaits is often referred to as the Zero
> Moment Point, or ZMP, method, because stability is enforced by ensuring
> that
> the center of pressure lies within the region of support at every instant
> during gait. Recently, researches have proposed a variety of methods to
> work around the need for straight knees while still using the ZMP
> criterion
> for gait generation.
>
> An alternative approach that uses straight knees during stance and seems
> more closely tied to biomechanics is passive dynamic walking as
> demonstrated
> by McGeer, Ruina, Kuo, and Wisse. This article in science news provided a
> good overview of the field:
> http://www.sciencenews.org/articles/20050806/bob8.asp
> Some videos of a powered passive dynamic walker, that make a good contrast
> to videos of asimo, from a different survey in Science:
> http://www.sciencemag.org/cgi/content/full/307/5712/1082/DC1
>
> Stuart Anderson
> PhD Candidate
> Robotics Institute
> Carnegie Mellon University
>
>
> On Dec 7, 2007 1:27 PM, Steven Abbott wrote:
>
>> Hello,
>>
>> I don't contribute that often, finding it far more interesting to just
>> "listen", but as its a friday afternoon!....
>>
>> I would think that by making asimo walk about like that it should do
>> something rather interesting to his centre of gravity, perhaps making it
>> slightly lower? The greater level of flexion in running would shift the
>> centre of gravity still further, which would make sense as running would
>> probably be even less stable.
>>
>> As to why, well, however human honda want to make asimo, his internal
>> bits
>> and pieces are going to inevitably have a very different weight
>> distribution to that of a biological human. It may be that he is slightly
>> top heavy, hence the need to shift the centre of gravity a bit when
>> walking?
>>
>>
>>
>> --
>> Steven Abbott
>> Researcher
>> Bioengineering Research Group
>> Anglia Ruskin University
>> 01245 493131 - ext 3317
>>
>> ---------------------------------------------------------------
>> Information about BIOMCH-L: http://www.Biomch-L.org
>> Archives: http://listserv.surfnet.nl/archives/Biomch-L.html
>> ---------------------------------------------------------------
>>
>
> ---------------------------------------------------------------
> Information about BIOMCH-L: http://www.Biomch-L.org
> Archives: http://listserv.surfnet.nl/archives/Biomch-L.html
> ---------------------------------------------------------------