Peter,
one of the distinctions will be to do with whether you define random
movement in terms of short or long range spatial scales. Imagine the
open space of a city with blocks say 50 or 100m long and streets say
10m wide. Now put some agents in that move in relatively short steps
(say 1 or 2m) in random directions. When they hit a wall the stop and
choose another direction at random within the open space. They will
buzz about and not get very far - that is they will take quite some
time to diffuse through the city system. Now think of them instead
taking relatively long random direction steps (say 200-500m) and again
stopping and re directing when they hit a wall. They will diffuse more
quickly, and if you work in an urban system with differentiated
morphology - main streets and back streets - my guess is that the
resulting distribution of path densities will correlate with both
space syntax global measures and observed pedestrian/traffic flows.
Alan
On 17 Nov 2009, at 10:56, Urban Lists Sust Urb wrote:
> Dear Bin,
>
> In my work, which is building upon spatial movement work undertaken by
> IVT at ETHZ, a critical factor in movement in the city is that it is
> not random. People have purpose and destination. They have certain
> spheres of influence within the city. Thus there is a set of people
> within any area who have purpose within that area, but may come from
> another ...
>
> I see what you say below in your response to Hans, but how do we
> translate the random generation to the purposeful use of a "world"
> network with destinations, where we are trying to understand the
> average overall purposeful behavior, as against average random
> behavior?
>
> Maybe. if there is a misunderstanding, it is because I am not an ABM
> modeler?
>
> Kind regards
> Peter
>
> On 17/11/2009, at 9:13 PM, Bin Jiang wrote:
>
>> Many thanks Hans for the detailed comments, my feedback to which is
>> as follows:
>> Skov-Petersen wrote:
>>> Hi Bin,
>>> A few comments/questions:
>>> 1) Apparently you are using two sw products: ArcGIS (for G�vle)
>>> and NetLogo (for London). Ir is not clear to me how the two were
>>> applied. I would assume that NetLogo was used for the agent
>>> simulation while ArcGIS was used for network analysis (metrics),
>>> but as it reads boh programmes were used for both purposes (but in
>>> different regions). Can you clear me up?
>> In the paper we just say a few words of the difference between the
>> two platforms. At the beginning, we put Gävle data in ArcGIS, while
>> London data in NetLogo. However, this is NOT essential. In other
>> words, to get the statistics there is no difference between the two,
>> but they do differ if visualization is concerned. Obviously in this
>> connection, NetLogo outperforms.
>>> 2) Is the NetLogo model-code you used for agents interaction with
>>> the network publicly available?
>> Yes.
>>> 3) are your purposely agents (II) applying a 'shortest path' search
>>> towards their target?
>> Yes, but shortest path can be computed in real time. It would reduce
>> the simulation speed significantly.
>>> 4) Wouldn't you assume that the purposely agents (II) could be
>>> further 'improved' by taking the probability (i.e. the number of
>>> potential facilities) into account when selecting targets/
>>> destinations as a probability weight 't application of a temporal
>>> dimension be considered. As it is, the simulation mimics the
>>> behaviour of taxies quite well (roaming short distance, any where,
>>> all during the day), but not the way e.g. home-work journeys will
>>> take place.
>> Of course as long as one has all these locations of potential
>> facilities. In our experiments, the destinations are randomly
>> generated, and they are randomly distributed.
>>> 5) One main finding - as I read it - is that ABM's are not required
>>> to simulate traffic flows (which is quite disappointing for an abm-
>>> modeler :-)).
>> What do you mean by this point? We relied on ABM for simulating
>> traffic flows as you can see.
>>> Nevertheless, your conclude that abm's provide us with new ways to
>>> study the rational behind human (spatial) behaviour, but do not
>>> further elaborate on this.
>> My point here is that drawn from our experiments ABMs provide a
>> means to study human spatial behavior instead of observing from the
>> real world.
>>> To me - and that is probably what you are saying - the thing is
>>> that the network (obviously) is the mandatory, bounding condition
>>> for transport behaviour. It is interesting (and efficient) to come
>>> up with indicators (metrics) that can predict human behaviour
>>> patterns (at a gross level), but that we need the agent-based
>>> approach to further enhance our behavioural understanding,
>>> especially when considering behaviour beyond the 'average being'.
>>> Right?
>> No, this is not what I intended to say. Also see above point. My
>> focus is understand average being rather than individual being.
>>
>> Cheers.
>>
>> Bin
>>> Hans
>>>
>>>
>>> >>> Bin Jiang <[log in to unmask]> 11-06-2008 17:00 >>>
>>> Hi, this paper might be of interest to space syntax researchers
>>> http://arxiv.org/ftp/arxiv/papers/0804/0804.1630.pdf
>>> any comments are very welcome.
>>>
>>> Cheers.
>>>
>>> Bin
>>>
>> --------------------------------------------------------
>> Bin Jiang
>> Division of Geomatics, KTH Research School
>> Department of Technology and Built Environment
>> University of Gävle, SE-801 76 Gävle, Sweden
>> Phone: +46-26-64 8901 Fax: +46-26-64 8828
>> Email: [log in to unmask] Web: http://fromto.hig.se/~bjg/
>> --------------------------------------------------------
>> European Associate Editor
>> Computers, Environment and Urban Systems: An International Journal
>>
>> NordGISci: http://fromto.hig.se/~bjg/NordGISci/
>> ICA Commission: http://fromto.hig.se/~bjg/ica/
Alan Penn
Dean of the Bartlett Faculty of the Built Environment,
Professor of Architectural and Urban Computing,
University College London,
Wates House, 22 Gordon Street,
London WC1H 0QB
[log in to unmask]
www.bartlett.ucl.ac.uk
www.vr.ucl.ac.uk
www.spacesyntax.com
|