Algorithms for Transportation

When people talk about algorithms the conversation tends to slant in the direction of SkyNet blowing up the world or high frequency trading bots crashing financial markets, these are not always the most practical applications for algorithms which can be anything from baking recipes to a multi-billion dollar search engines. However knowing what they are is only the first step, we are here to find what the value of algorithms are, specifically for the transportation industry. To answer this we went through a large collection of commentary all about algorithms by experts in the field, gathered by Pew Research Center. These experts provided some great quotes on how algorithms help save time, money and potentially the world itself.

If you transport people with a vehicle you are going to have routes, if you have multiple vehicles doing multiple routes you have a problem. This is called the Vehicle Routing Problem or VRP. It is a subset of the Travelling Salesman Problem. These problems both involve finding the ideal route between a number of destinations. The challenge with the VRP is the number of possible solutions. Even for a modest sized fleet of vehicles, let’s say ten vans, doing 100 trips in a day, there would be trillions of ways to route those vehicles. If you want an optimal solution, one that uses the fewest vehicles in the shortest amount of time, you would need to build and test each solution, this would take an impossible amount of time, even for a simple problem and it is like finding a needle in a mountain sized haystack. And as soon as something changes, such as an additional trip or cancellation, a plan is no longer optimal. This is where algorithms come in. Chris Kutarna, fellow at the Oxford Martin School, explains, “algorithms are an explicit form of heuristic, a way of routinizing certain choices and decisions so that we are not constantly drinking from a fire hydrant of sensory inputs.”

These sensory inputs are becoming bigger everyday. First it was just GPS and map data, soon it will be dedicated sensor suites covering every car, as claimed by the Financial Times: “Data is the new oil,” Consider self-driving cars: between cameras (collecting 20 and 40 megabytes per second), sonar (10 to 100 KB/sec) and Lidar (10 to 70 MB/sec), Intel estimates there will be 4,000 gigabytes of data everyday from each car, multiply that by 300 million cars on North American streets, and you’ve got an enormous amount of information to process.

That “fire hydrant” of inputs is all the available information for making decisions. Just to build routes, not even thinking of processing sensor data, you must consider the location of riders, their destinations, the locations and capacity of the vehicle, and the possible roads available. Combine these with other constraints like the fixed schedules of other vehicles, add the fact that all of these factors are dynamically changing and that planning must be done not just for one trip, but for all possible trips throughout the day. This is why the VRP is so challenging, and why we need algorithms to help.

As Demian Perry, director of mobile at NPR, explains algorithms are “helpmates” which add efficiency, “an algorithm is just a way to apply decision-making at scale… mass-produced decisions are, if nothing else, more consistent. Depending on the algorithm (and whom you ask), that consistency is either less nuanced or more disciplined than you might expect from a human.” Let’s break these points down, as said before there are usually trillions of possible ways to solve a vehicle routing problem, and those types of decisions will either save or cost a lot of money. By using an algorithm to consistently make decision based on data that is consistently fed into it, using sensors and other technologies, you will know that your problem is being consistently solved. For people transporters that means less time on the road, less gas being burned and happier riders who get reliable service. As Robert Atkinson, president of the Information Technology and Innovation Foundation, says, “Like virtually all past technologies, algorithms will create value and cut costs, far in excess of any costs. Moreover, as organizations and society get more experience with use of algorithms there will be natural forces toward improvement and limiting any potential problems.”

From non-emergency medical transportation to public transit to ride sharing and future autonomous vehicle fleetsdecisions and data are made everyday on such a scale that a human mind can not optimally find solutions, especially when they need to be made in real-time, not even most algorithms can handle those types of problems without the right amount of time for calculating. Furthermore, while there might be many versions of these types of algorithms buried away in PhD projects and university libraries, to work, they need to be connected to a surrounding infrastructure to have good inputs and outputs of data. That requires reliable connections with vehicles, riders and operators. Algorithms working within the transportation industry also need to have the right fleet, one that allows flexible, dynamic routes and has high ridership demand and high capacity of seats. Those factors will allow the right algorithm to extract the maximum amount of value from a fleet of vehicles everyday. As reliable as a recipe bakes a cake with the right user, an algorithm can do the same with an entire transportation system. Do you have know of a fleet like that, or another transportation use case that needs the help of algorithms? Let us know at info@pantonium.com.

 

Image Credit: x6e38
2017-04-14T16:43:51+00:00 April 14th, 2017|Non-Emergency Medical Transportation, Pantonium, Technology|Comments Off on Algorithms for Transportation