external image Uber1.jpg


Andre King & Andrew Do


Currently road networks are not optimally used by society as a whole. People drive single-passenger vehicles to get to and from work routinely every day under the guise of flexibility in travel. This leads to heavy congestion in the system, which, in turn leads to more time spent on average by each driver. As the population increases, the modus operandi is neither sustainable nor scalable---more people would just mean more cars on the road, further exacerbating the problem. Moreover, the number of cars on the road is directly correlated with greenhouse emissions, so any reducing the number of drivers on the road every day would simultaneously address environmental concerns.

external image carpool-lane.jpg


In established urban centers, the infrastructure is difficult to change, so the more natural solution is to change what people are doing. We believe that
people would change their commuting habits if an easy, flexible, and reasonably-priced platform for carpooling were made available. Our initial focus will be on dense urban centers with large commercial and industrial zones surrounded by suburbs. The key idea is that while the probability that neighbors will be commuting to the exact same location is very low, the number of people with long overlapping commute paths is very high. In other words, if we relax the condition for exact matches (as in traditional carpools), then we can easily find people who are leaving from the same part of town and are headed to clustered destinations.

external image carpool.png


The Uber platform needs to be equipped with a carpool scheduler and optimizer. The most important pieces of information are people's work and home locations and their usual departure times. The scheduler matches batches of people who are headed from and to similar sections of town with an Uber driver who will be given an optimized route for pick-up and drop-off. Priority will be given to people who schedule ahead of time, and drivers will be better-compensated for committing to giving rides at rush hour times.

We then cluster companies based on radius and ease-of-access to each other and approach them with a "Green Alliance" model: they aggressively advertise internally to their employees in return for a share of the profits generated by their employees. In addition, membership will then show their commitment to environmental stewardship.

external image banner_slide6.png


We measure the success of our technology by looking at user return rates. If they are using Uber-To-Work at least once a week, we can go to calculate how much carbon emission is prevented by our users. We can also take measurements on traffic rates in areas to see the effect that an increase in carpooling has on traffic conditions.


In the long term, large user adoption in geographic areas will lead to faster travel time to and from work due to a decrease in traffic. In the short term, we evaluate our experiment by seeing how quickly we can get people to and from work. If we can make sure that all users take the same amount of time to get to work as before, we will have guaranteed improved conditions for the environment, because there will be less cars on the road, and for people, because they have extra time will riding in the carpool to do more work or talk to new people.

What new data would you use? How would you obtain it? How would you use it?

We would need to ask for user's home and work locations, which we can get by asking users when they register with Uber-To-Work. We would use it to cluster users into groups that would carpool to work together.

How will this new data improve the end user experience? Clarify how you can really improve the product.

User's will prefer carpooling with Uber to work because they will feel good for contributing to an environmental cause, and they can have someone driving them to work which frees up time for them.

external image blogpost2-carpool.jpg?w=1200