A Dynamic Traveling Salesman Problem with Stochastic Arc Costs

Toriello, Alejandro; Haskell, William B.; Poremba, M.

doi:10.1287/opre.2014.1301

Cited by 45 publications

(20 citation statements)

References 66 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We do not include the ALP lower bound, as our preliminary experiments revealed it to be weaker than the PIR bound. Similar behavior has been observed in other stochastic routing contexts, e.g., [53].…”

Section: Computational Experimentssupporting

confidence: 85%

“…Different stochastic and dynamic VRPs focus on uncertainty in different sets of parameters. Some examples are the VRP with stochastic demands [2,11,29,43,49], the VRP with stochastic travel times [18,35,36,39,40,41,51,53], and the VRP with probabilistic customers [8,15,24,32,33,37,55].…”

Section: Vehicle Routing and Dispatch Problemsmentioning

confidence: 99%

“…One widely used ADP method is to develop approximations for the optimal cost-to-go function and use it to select an approximately optimal action at any encountered state; see [9,45]. Rollout algorithms [10] have been widely applied for stochastic routing models, e.g., [29,49,53].…”

Section: Mdp and Approximate Dynamic Programmingmentioning

confidence: 99%

See 2 more Smart Citations

The One-Dimensional Dynamic Dispatch Waves Problem

Klapp

Erera

Toriello

2018

Transportation Science

Self Cite

View full text Add to dashboard Cite

We study same-day delivery (SDD) systems by formulating the Dynamic Dispatch Waves Problem (DDWP), which models a depot where delivery requests arrive dynamically throughout a service day. At any dispatch epoch (wave), the information available to the decision maker is (1) a set of known, open requests which remain unfulfilled, and (2) a set of potential requests that may arrive later in the service day. At each wave, the decision maker decides whether or not to dispatch a vehicle, and if so, which subset of open requests to serve, with the objective of minimizing expected vehicle operating costs and penalties for unserved requests. We consider the DDWP with a single delivery vehicle and request destinations on a line, where vehicle operating times and costs depend only on the distance between points. We propose an efficient dynamic programming approach for the deterministic variant, and leverage it to design an optimal a priori policy with predetermined routes for the stochastic case. We then show that fully dynamic policies may perform arbitrarily better than a priori ones, and propose heuristics and dual bounds for this case.One enhancement to customer service in this sector is same-day delivery (SDD). Several e-retailers and logistics service providers have introduced programs in major US cities; see Table 1. We define SDD as a distribution service where consumers place orders on the same day that they should be delivered. For companies that implement SDD, it is imperative to both offer high levels of customer service and keep logistics costs as low as possible.Providing SDD services requires two core logistics processes: (1) order management at the stocking location, including receiving, picking, and packing orders; and (2) order distribution from the stocking location to delivery locations. To date, two classes of service providers have deployed SDD services: retailers offering items primarily from owned stocks (in distribution centers or retail stores) such as Amazon and Walmart, and logistics providers serving as intermediaries that pick up packages from stocking locations and deliver them to customers, such as Google and Instacart as well as USPS, FedEx and UPS. Retailers must manage both core logistics processes, while logistics providers are typically concerned with the second one. This research effort is a study of primary tradeoffs in SDD distribution. We formulate the Dynamic Dispatch Waves Problem (DDWP) as a Markov Decision Process (MDP). The DDWP models the dynamics of a single dispatch facility (depot) where customer order requests arrive dynamically throughout an operating day. At any decision epoch, which we call a wave, the logistics operator maintains a set of known open requests with known delivery destinations and a set of potential requests that may arrive before the end of the day. At each wave, the operator decides whether or not to dispatch vehicles loaded with known orders, and the vehicle routes for dispatched orders. The objective is to minimize expected operational costs and expect...

show abstract

Section: Computational Experimentssupporting

confidence: 85%

Section: Vehicle Routing and Dispatch Problemsmentioning

confidence: 99%

See 1 more Smart Citation

The One-Dimensional Dynamic Dispatch Waves Problem

Klapp

Erera

Toriello

2018

Transportation Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…With limited scenarios generated, however, the true optimal solution might not be achieved by this sampling method. A dynamic TSP is considered by [17] where arc costs are unknown ahead of time except its distribution but are revealed dynamically before the decision is executed. An approximate linear programming model was employed to help a salesman choose his next destination.…”

Section: Introductionmentioning

confidence: 99%

Decomposition-based exact algorithms for risk-constrained traveling salesman problems with discrete random arc costs

Huang

Zheng

2014

Optim Lett

View full text Add to dashboard Cite

Recently increasing attentions have been given to uncertainty handling in network optimization research. Along this trend, this paper discusses traveling salesman problem with discrete random arc costs while incorporating risk constraints. Minimizing expected total cost might not be enough because total costs of some realizations of the random arc costs might exceed the resource limit. To this respect, this paper presents a model of the traveling salesman problem that incorporates risk constraints based on Conditional Value at Risk to evaluate those worst-cost scenarios. Exact solution methods are developed and applied on the risk-constrained traveling salesman problem. Numerical experiments are conducted, and the results show the ability of the proposed methods in reducing the computational complexity.

show abstract

“…While literature on the stochastic TSP with profits and time constraints mainly focuses on developing a priori routes, dynamic solutions have been extensively developed for vehicle routing problems (VRPs). Our work is similar to the literature in using approximate dynamic programming (ADP) to solve routing problems, especially in developing heuristic routing policies via rollout procedures (Toriello et al (2014), Secomandi (2000, Novoa and Storer (2009), Goodson et al (2013), and Goodson et al (2015)). To approximate the cost-to-go of future states, Secomandi (2000Secomandi ( , 2001, Novoa and Storer (2009), and Goodson et al (2013) use a priori routes as heuristic policies, while Toriello et al (2014) apply the approximate linear programming (ALP).…”

Section: Approximate Dynamic Programmingmentioning

confidence: 80%

Stochastic orienteering on a network of queues with time windows

Zhang¹

View full text Add to dashboard Cite

A Dynamic Traveling Salesman Problem with Stochastic Arc Costs

Cited by 45 publications

References 66 publications

The One-Dimensional Dynamic Dispatch Waves Problem

The One-Dimensional Dynamic Dispatch Waves Problem

Decomposition-based exact algorithms for risk-constrained traveling salesman problems with discrete random arc costs

Stochastic orienteering on a network of queues with time windows

Contact Info

Product

Resources

About