Highly Efficient and Scalable Multi-hop Ride-sharing

Abstract

On-demand ride-sharing services such as Uber and Lyft have gained tremendous popularity over the past decade, largely driven by the omnipresence of mobile devices. Ride-sharing services can provide economic and environmental benefits such as reducing traffic congestion and vehicle emissions. Multi-hop ride-sharing enables passengers to transfer between vehicles within a single trip, which significantly extends the benefits of ride-sharing and provides ride opportunities that are not possible otherwise. Despite its advantages, offering real-time multi-hop ride-sharing services at large scale is a challenging computational task due to the large combination of vehicles and passenger transfer points. To address these challenges, we propose exact and approximation algorithms that are scalable and achieve real-time responses for highly dynamic ride-sharing scenarios in large metropolitan areas. Our experiments on real-world datasets show the benefits of multi-hop ride-sharing services and demonstrate that our proposed algorithms are more than two orders of magnitude faster than the state-of-the-art. Our approximation algorithms offer a comparable trip quality to our exact algorithm, while improving the ride-sharing request matching time by another order of magnitude.

Preferential Optimization of Multi-Hop Dynamic Ridesharing Based on R-Trees and Multi-Agent Systems

“…In this context, a spatial indexing technique known as R-tree makes it possible to index objects dynamically in a tree and to optimize zone or node overlapping queries (11)(12)(13). The R-tree has been used for taxi ridesharing and has proven to be very efficient (6,14,15). In this paper, we propose to extend the scope of use of R-trees to the private driver ridesharing problem.…”

“…The Aggregate Nearest Neighbor (ANN) Problem. Previous works have pointed out the parallel between the road networks in graph form (25,28), the multi-hop dynamic ridesharing problem (14) and the Group or Aggregate Nearest Neighbor problem (first called GNN and later ANN) (24). If we consider T as the set of available transfer nodes and N as the set of users starting and ending nodes, the ANN (aggregate nearest neighbor) problem can be formulated as follows : an ANN query retrieves the point(s) of T with the smallest sum of distances to all points in N. When looking for the optimal transfer node, we are looking for the node that minimizes the sum of the detour distances associated with the path sharing.…”

“…With the advent of multi-hop ridesharing and to obtain an exact and multi-objective solution, the research is oriented toward an approach of the ridesharing problem via the group nearest neighbor (GNN) or aggregate nearest neighbor (ANN) problem ( 24 ). Indeed, since multi-hop implies the search for the optimal transfer node, heuristics based on nearest neighbor search algorithms have been developed and have been enhanced to be more scalable ( 14 , 25 , 26 ). It is from this perspective that we propose to extend the incremental Euclidean restriction (IER) reference algorithm to jointly evaluate multi-hop and single-hop solutions in a multi-objective way ( 25 ).…”

“…To do so, they maintain a trip schedule consisting of different tables storing spatio-temporal information. In this context, we extend the trip scheduling architecture presented in Tong et al and in two works by Xu et al by adding several new arrays ( 6 , 14 , 15 ). The above previous works’ architecture mainly describes three arrays: “ arr array” stores the shortest time needed to reach each stop, “ ddl array” stores the latest possible arrival time (deadline) of each stop to respect the constraints of the driver (latest arrival time) or the riders (waiting time and latest arrival time) carried, and “ slk array” stores the maximum available detour time before each stops (slack time).…”

Sub-trajectory Similarity Join with Obfuscation

“…Such partially similar trajectories are gaining importance in applications such as contact tracing for managing epidemics, e.g., to find people in close contact with confirmed cases of COVID-19 for a duration of over 15 minutes 1 . Another example is to compute partially similar trajectories to find matches to form multi-hop goods delivery or car-pooling arrangements that allow transits [31].…”