Smartphone data from Deliveroo riders delivering meals for restaurant-to-home shows that bicycles are faster than cars. In towns and cities, bicyclists are also often faster than motorized two-wheelers.
Delivering millions of orders from thousands of restaurants to consumers with self-employed couriers equipped with smartphones is a complex vehicle routing problem. Consumers want hot food within 30 minutes, restaurants want meals picked up when cooked, riders want multiple deliveries per hour, and Deliveroo needs to make money.
Carlton Reid reports about in on Forbes after talking with Deliveroo co-founder Will Shu. When Shu started, Shu didn’t have the benefit of a bespoke routing algorithm. Deliveroo initially used first in, first out queueing methodology. But Shu’s experience as a rider made him appreciate that optimizing the many parts of the ordering and delivery process would be key to Deliveroo’s success. And he also figured out that some jobs would be better suited to cyclists and others to motorists.
Today, Shu’s routing guesswork has been replaced by machine-learning technology. Frank is the name for Deliveroo routing algorithm. Frank predicts the timings of every order depending on the meal ordered, the location of the restaurant, the time of day and the day of the week, the number of riders logged in to courier app, and the distance from the restaurant to the customer.
Carlon Reid concludes: delivering by bike is faster than by car. In one way streets, bypassing roundabouts and no hinder from congestion. Circling to find a free parking space adds precious time; time that cyclists use to pedal to their destinations.
Source: Carlton Reid in Forbes