Saving emissions with sustainable routing
For over a decade, Uber’s platform has connected drivers who have cars with riders who need to get from point A to point B. To help facilitate these trips, the Uber Earner app suggests what we think is the best route for the driver to follow. But what is the “best” route from point A to B? Historically, we have been treating the best routes as those with a minimal combination of distance and duration. With the effects of climate change becoming more obvious every day and the ability of small changes across billions of trips to have significant impact, we recently decided to take into account one new factor — sustainability.
We worked with the National Renewable Energy Laboratory (NREL) to estimate fuel consumption and associated CO₂ emissions for each route we suggest to drivers. The NREL RouteE model allows us to estimate the granular fuel consumption for each road segment in the United States. We consider road and driving characteristics, such as gradients and average speed for each road segment to estimate the most accurate fuel consumption for each route. With this detailed data, we are able to quantify estimated fuel consumption among possible routes and show drivers which routes are more sustainable (see the product image below).
We went further to made changes to our routing algorithm based on the estimated fuel consumption data. We hypothesized that shorter routes lead to lower fuel consumption, so we set up an experiment: While keeping the trip duration (in other words, estimated arrival time, ETA) intact as much as possible, we were able to tweak our routing algorithm to reduce distance for some trips. We then used the NREL model to estimate the difference in fuel consumption. During this experiment, the average estimated fuel consumption for the trips in the treatment group was 0.65% lower compared to the trips in the control group where the original algorithm was used. The decrease in trip distance is the main lever causing this reduction of fuel consumption. Notably, the change in trip duration on the alternative route was not statistically significant. This means we found a new “sweet spot” on the complex tradeoffs among all factors we use for the routing algorithm– distance, duration, and sustainability.
Following the positive outcome of this experiment, we launched this new routing algorithm to all Uber mobility trips in the United States. Assuming the same level of improvement in fuel consumption for trips served by non electric vehicles, we estimate that so far we have reduced fuel consumption by 1.1 million gallons, which is equivalent to more than 9,800 metric tons of CO₂ emissions¹. We expect this impact to accumulate more and more every day, every year as drivers complete their trips with our new routing algorithm.
However, we aren’t stopping here. Our future plan includes more direct integration of the model to estimate fuel consumption into the routing algorithms. We also believe it’s important to consider vehicle-specific details to estimate more accurate fuel consumption. In addition to that, we would like to eventually expand the new eco-routing algorithm to other geographies, as we operate in 70 countries on 6 continents. Our sustainability journey continues until we achieve our commitment to have zero tailpipe emissions for all vehicles on the platform. Don’t stop til zero!
¹ Assume trips are served by 30 MPG internal combustion vehicles, excluding trips served by fully electric vehicles.
