By Jeff Hulett, May 27, 2017
Situation:
Sitting in traffic consumes a significant amount of time. In major cities and before the pandemic, the average commuting time can be over 1 hour per day, (CNN, March 3, 2016) with many drivers in major cities having significantly longer daily commutes. To help people make more informed route decisions, technology has been evolving to assist people in making route decisions. Popular technology companies include Waze, Microsoft Maps, and others. The evolution of traffic route assistance has evolved from a one-way route map plan (One way from the satellite to the end-user with central map route information) to more sophisticated 2-way communication, where user-based information (traffic, police, traffic cameras, accidents, etc.) are communicated to the satellite and combined with the central map route information. The decision about the route is generally based on a constraint optimization algorithm with a linear programming basis (“algo”), which takes into account constraints like estimated route time to deliver the optimal route. More recent technology allows dynamic updating where new user information can cause the algo to update based on new user-provided information. (e.g., an accident occurs and the algo recalculates because the previous route chosen no longer is the fastest route to the destination.)
Complication:
Route decisions can be more complex than the current algos allow. An individual’s route decision will likely encompass other behavioral factors, such as:
Cost of the route – Current algos does not generally handle cost inputs other than restricting a toll-based route. Today, express lanes are becoming more common. In effect, express lanes allow the user to “buy” their way out of traffic, trading money for time. The cost of the express lane changes, based on current demand. (i.e., the higher cost during rush hour) Given the amount of data technology companies have available based on actual routes and that costs are publically disclosed, it is assumed the technology companies have the technical ability to dynamically calculate route cost.
Risk of the route – while the current algo can solve the route optimization problem based on a route time optimization, they do not solve the risk the individual routes have to change (lengthen) while the user is en route. That is, what is the probability the time to destination will increase or decrease more than a particular standard (say, 10%). Given the amount of data technology companies have available based on actual routes and time performance of those routes at particular times of day (say, rush hour) it is assumed the technology companies have the technical ability to calculate route risk.
Resolution:
When a user inputs their route, there are some simple behavioral questions that could be answered for every route or a default set of behavioral answers that are preprogrammed based on the user set up and are offered to be changed for the particular route. Potential behavioral questions include:
How important is it that you get to your destination based on this ETA? (Must, Very Important, Average Importance, On a Sunday Drive)
**”This ETA” would be some time above the optimal route time that allows for a lower risk route. This question is to help understand the route risk profile of the user.
How much is your time worth right now? (I will pay almost anything for an extra minute of time, $100/hour, $10 / hour, enjoying the scenery)
**”I would pay almost anything” would be for an emergency situation. This question is to help understand the route cost desire of the user.
When these 2 behavioral questions are answered, the response becomes an input to the algo and impacts the weight given to the route recommendation. The recommended route could be changed based on the answer to these questions.
Use case examples:
Example 1: A person is going to work during rush hour. They are in a big hurry as they have a meeting with their boss and they cannot be late. They must be there no later than 45 minutes. The route options include
30 minutes with a route risk equivalent to 20 minutes.
35 minutes with a route risk equivalent to 5 minutes.
If the user answered the route risk question as “Must” the algo would recommend route 2. If no answer, the algo would default to the fastest route, recommending route 1.
Example 2: A person is going to work and is anxious to get working on an important project. They believe their time is worth $100 / hour. The route options include:
35 minutes, cost of $10 on the express lane
1 hour, no cost
The incremental cost is equivalent to $24 / hour, which is < $100 user time value, thus algo will recommend route 1.
Example 3: A person is going to work and is anxious to get working on an important project. They believe their time is worth $10 / hour. The route options include:
35 minutes, cost of $10 on the express lane
1 hour, no cost
The incremental cost is equivalent to $24 / hour, which is > $10 user time value, thus algo will recommend route 2.
Example 4: A person is driving their child to a championship soccer game. They need to be there in 45 minutes. The person believes it is very important the child be there on time and has the resources to pay for the time. The route options include:
35 minutes, cost of $10 on the express lane
1 hour, no cost
If the user answered the route cost question “Must” the algo would recommend route 1.
Example 5: A person is driving their child to the store to clothes shop. They have a little time to talk and the user is looking forward to talking to their child on the way. They would like to be there in 45 minutes but have flexibility. The route options include:
35 minutes, cost of $10 on the express lane
1 hour, no cost
If the user answered the route cost question “enjoying the scenery” the algo would recommend route 2.
Comentários