Neural approximate dynamic programming for on-demand ride-pooling

On-demand ride-pooling (e.g., UberPool, LyftLine, GrabShare) has recently become popular because of its ability to lower costs for passengers while simultaneously increasing revenue for drivers and aggregation companies (e.g., Uber). Unlike in Taxi on Demand (ToD) services – where a vehicle is assig...

Full description

Saved in:
Bibliographic Details
Main Authors: SHAH, Sanket, LOWALEKAR, Meghna, VARAKANTHAM, Pradeep
Format: text
Language:English
Published: Institutional Knowledge at Singapore Management University 2020
Subjects:
Online Access:https://ink.library.smu.edu.sg/sis_research/5992
https://ink.library.smu.edu.sg/context/sis_research/article/6995/viewcontent/5388_Article_Text_8613_1_10_20200511.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Singapore Management University
Language: English
id sg-smu-ink.sis_research-6995
record_format dspace
spelling sg-smu-ink.sis_research-69952021-06-08T09:21:58Z Neural approximate dynamic programming for on-demand ride-pooling SHAH, Sanket LOWALEKAR, Meghna VARAKANTHAM, Pradeep On-demand ride-pooling (e.g., UberPool, LyftLine, GrabShare) has recently become popular because of its ability to lower costs for passengers while simultaneously increasing revenue for drivers and aggregation companies (e.g., Uber). Unlike in Taxi on Demand (ToD) services – where a vehicle is assigned one passenger at a time – in on-demand ride-pooling, each vehicle must simultaneously serve multiple passengers with heterogeneous origin and destination pairs without violating any quality constraints. To ensure near real-time response, existing solutions to the real-time ride-pooling problem are myopic in that they optimise the objective (e.g., maximise the number of passengers served) for the current time step without considering the effect such an assignment could have on assignments in future time steps. However, considering the future effects of an assignment that also has to consider what combinations of passenger requests can be assigned to vehicles adds a layer of combinatorial complexity to the already challenging problem of considering future effects in the ToD case. A popular approach that addresses the limitations of myopic assignments in ToD problems is Approximate Dynamic Programming (ADP). Existing ADP methods for ToD can only handle Linear Program (LP) based assignments, however, as the value update relies on dual values from the LP. The assignment problem in ride pooling requires an Integer Linear Program (ILP) that has bad LP relaxations. Therefore, our key technical contribution is in providing a general ADP method that can learn from the ILP based assignment found in ride-pooling. Additionally, we handle the extra combinatorial complexity from combinations of passenger requests by using a Neural Network based approximate value function and show a connection to Deep Reinforcement Learning that allows us to learn this value-function with increased stability and sample-efficiency. We show that our approach easily outperforms leading approaches for on-demand ride-pooling on a real-world dataset by up to 16%, a significant improvement in city-scale transportation problems. 2020-02-01T08:00:00Z text application/pdf https://ink.library.smu.edu.sg/sis_research/5992 info:doi/10.1609/aaai.v34i01.5388 https://ink.library.smu.edu.sg/context/sis_research/article/6995/viewcontent/5388_Article_Text_8613_1_10_20200511.pdf http://creativecommons.org/licenses/by-nc-nd/4.0/ Research Collection School Of Computing and Information Systems eng Institutional Knowledge at Singapore Management University Approximate dynamic programming Approximate value function Assignment problems Combinatorial complexity Integer linear programs Origin and destinations Transportation problem Numerical Analysis and Scientific Computing Operations Research, Systems Engineering and Industrial Engineering Transportation
institution Singapore Management University
building SMU Libraries
continent Asia
country Singapore
Singapore
content_provider SMU Libraries
collection InK@SMU
language English
topic Approximate dynamic programming
Approximate value function
Assignment problems
Combinatorial complexity
Integer linear programs
Origin and destinations
Transportation problem
Numerical Analysis and Scientific Computing
Operations Research, Systems Engineering and Industrial Engineering
Transportation
spellingShingle Approximate dynamic programming
Approximate value function
Assignment problems
Combinatorial complexity
Integer linear programs
Origin and destinations
Transportation problem
Numerical Analysis and Scientific Computing
Operations Research, Systems Engineering and Industrial Engineering
Transportation
SHAH, Sanket
LOWALEKAR, Meghna
VARAKANTHAM, Pradeep
Neural approximate dynamic programming for on-demand ride-pooling
description On-demand ride-pooling (e.g., UberPool, LyftLine, GrabShare) has recently become popular because of its ability to lower costs for passengers while simultaneously increasing revenue for drivers and aggregation companies (e.g., Uber). Unlike in Taxi on Demand (ToD) services – where a vehicle is assigned one passenger at a time – in on-demand ride-pooling, each vehicle must simultaneously serve multiple passengers with heterogeneous origin and destination pairs without violating any quality constraints. To ensure near real-time response, existing solutions to the real-time ride-pooling problem are myopic in that they optimise the objective (e.g., maximise the number of passengers served) for the current time step without considering the effect such an assignment could have on assignments in future time steps. However, considering the future effects of an assignment that also has to consider what combinations of passenger requests can be assigned to vehicles adds a layer of combinatorial complexity to the already challenging problem of considering future effects in the ToD case. A popular approach that addresses the limitations of myopic assignments in ToD problems is Approximate Dynamic Programming (ADP). Existing ADP methods for ToD can only handle Linear Program (LP) based assignments, however, as the value update relies on dual values from the LP. The assignment problem in ride pooling requires an Integer Linear Program (ILP) that has bad LP relaxations. Therefore, our key technical contribution is in providing a general ADP method that can learn from the ILP based assignment found in ride-pooling. Additionally, we handle the extra combinatorial complexity from combinations of passenger requests by using a Neural Network based approximate value function and show a connection to Deep Reinforcement Learning that allows us to learn this value-function with increased stability and sample-efficiency. We show that our approach easily outperforms leading approaches for on-demand ride-pooling on a real-world dataset by up to 16%, a significant improvement in city-scale transportation problems.
format text
author SHAH, Sanket
LOWALEKAR, Meghna
VARAKANTHAM, Pradeep
author_facet SHAH, Sanket
LOWALEKAR, Meghna
VARAKANTHAM, Pradeep
author_sort SHAH, Sanket
title Neural approximate dynamic programming for on-demand ride-pooling
title_short Neural approximate dynamic programming for on-demand ride-pooling
title_full Neural approximate dynamic programming for on-demand ride-pooling
title_fullStr Neural approximate dynamic programming for on-demand ride-pooling
title_full_unstemmed Neural approximate dynamic programming for on-demand ride-pooling
title_sort neural approximate dynamic programming for on-demand ride-pooling
publisher Institutional Knowledge at Singapore Management University
publishDate 2020
url https://ink.library.smu.edu.sg/sis_research/5992
https://ink.library.smu.edu.sg/context/sis_research/article/6995/viewcontent/5388_Article_Text_8613_1_10_20200511.pdf
_version_ 1770575730027003904