Category Archives: Cloudcommuting

Cloudcommuting 02 : Experiments in Voronoi Mapping

Screen Shot 2014-09-16 at 9.51.25 PM

Screen Shot 2014-09-16 at 9.50.19 PM

Composite

The above images show a snap shots of dock availability and bike availability throughout the CitiBike system. The image in magenta shows bike availability at 9:30pm on Tuesday. The image below it, in cyan, shows Dock Availability at 9:30pm on Tuesday. Finally, I composited the two images to show their relationship. 

The image was generated with Processing, this is how it works:

I based my code on Dimitris Papanikolaou’s example sketch [insert gitHub link]. The Processing sketch took in JSON (from CitiBike’s website) and then parsed each JSON field into variables that Processing could work with. The variables (such as latitude, longitude and dock availability)  were then passed to a Voronoi class (from Toxiclibs) to generate the cells. In the end, I overlaid an SVG that I made in Illustrator to give the data a little context. Below is the image without the SVG overlay.

Screen Shot 2014-09-16 at 9.41.30 PM

gitHub repo is here.

Cloudcommuting : Assignment 01 Reading Response

In ME++ Mitchell talks about “electronic nervous systems” of intelligent urban environments. Discuss an example of an intelligent urban system you are familiar with and discuss the elements of the feedback loop, how its form of governance works, and who are its stakeholders (goals, decision makers, evaluators, etc.).

The concept of an intelligent urban system is quite broad, varied, and encompasses many types of systems; cctv, mass transit, emergency response, private security systems, community websites, delivery and couriers, government, infrastructure, waste, roadways, etc. With such a variety of typologies intelligence quickly becomes a highly relative term. Of these systems, one in which I’m fairly familiar with is a private vehicle sharing program called ZipCar.

I have been using ZipCar for about six years now. It’s a relatively intelligent urban transportation system that provides a variety of vehicles for on-demand use. As a user, I reserve a car via a web or mobile interface. My account is linked to a credit card, and I am charged an annual membership fee as well as a per-hour usage fee for my time in the car. I do not pay for gas or insurance. I retrieve the car from one of many set locations, drive it, and return it to the same location within my allotted reservation window. The system is governed electronically through a combination of human-computer interfaces (me and my phone or web browser) and an RFID card reader that is mounted to the windshield of the car that will either grant or deny access to the vehicle. The intelligence largely lives in a database of users, reservations, vehicles, and locations. Gas is paid for by credit card that lives in the vehicle. When there are issues, a human is available via telephone to resolve the issues. I list these systems to highlight the conglomerate of systems that are tied into ZipCar’s operation; web enabled databases, radio frequencies, humans, global financial institutions providing credit, roadways, and architecture (in some cases).

The stakeholders in the systems are both private and public, but mostly private. Its primary goal, as a company is to make tons of money, and it has continually been heading in that direction as it aggressively expands to new markets. The public can be seen as a stakeholder in ZipCar’s system as well, if only in a minor way; there is an incentive to reducing the amount of cars in urban environments. A reduction means less space dedicated to idle vehicles.

Compare different models of sharing that exist (or you can think of) in MoD systems (e.g. vehicle sharing, parking sharing, ride sharing, etc.) What operation/control problems do they have? What would the ideal form of sharing be for you and how would its resources be controlled?

The single largest operation and control problem that MoD systems have is resource management and system balancing. I believe that in a large part this is due to the fact that their systems exists in much larger ecosystem of systems as well as human demands influenced by environmental constraints.

For instance, with a car sharing system like ZipCar, the reliability of the system is variable and based on factors such as traffic, human error, mechanical failure, and maintenance. One advantage that a system like ZipCar has over other systems is that the resource (the car) is always returned to it’s origin which means relatively little rebalancing of the system. The system also has variable rates; weekends are more expensive to drive due to high demand, and more spacious or luxurious vehicles carry a higher hourly rate. 

Bike share systems are faced with much larger problems due the way the resources are used. Bikes are generally used asymmetrically, meaning they aren’t always returned to their point of origin. There is also a flat fee for use (with the exception of overtime penalties) which means that incentivizing the system for self-balancing becomes a difficult feat. There have been countless times when I’ve used a bike off-peak and spent up to a half hour trying to find an available (or functional) bike or an empty dock, which dissuades me from using the system all together.

The ideal system for me is a bike share that takes a few lessons from ZipCar. ZipCar doesn’t share quite the same system stress as a bike share, but it’s still there; there are peak usage times when it is maddeningly difficult to find a vehicle within a reasonable distance. However, I believe having a variable pricing structure opens up zipcar to incentivizing itself to regulate its system. If a bike share like CitiBike could adjust itself and its pricing structure or form an alliance with local businesses to incentivize use, I think some of its operational stress would be reduced. For instance, if riders are prompted with a discount at a coffee shop or grocery store that is closer to a more balanced station, perhaps they will help regulate the system.

Cloudcommuting : Midterm Proposal

Saki, Dan, and I would like to discuss two proposals for analysis and simulation.

Analysis 1: How does the citiBike infrastructure interface with public transit systems?

We will study and analyze the relationship of citiBike stations with other mass transit nodes. Do riders use multiple systems in tandem? Are there financial incentives to compete or cooperate with other modes of mass transit? We will use our analysis to simulate and optimize station locations.

Analysis 2: Who is using the citiBike system?

We will study and analyze the users of the citiBike system. What user groups are habitual users? How do user identities inform the growth of the citiBike infrastructure? We will create a parametric simulation of the system growth.

photo