|
Saeedeh Dehghani Firoozabadi Université du Québec à Montréal |
When was the last time you wanted to send a parcel to another country? For me, it was a couple of months ago when I tried to send a birthday gift to my friend. Like always, I checked post office prices. They were crazy, and the delivery times were too long. Thoroughly disappointed, I googled for a cheap and fast miracle! I found some websites where you can find people traveling along the same route who can transport your parcel. Wow, that’s amazing! There is always a traveler, which is much cheaper than post office prices. I do believe it should be greener. Overall, it was an excellent idea. Do we have it in other types of deliveries, like urban deliveries?
I continued reading about this and found similar websites for urban parcel deliveries and daily grocery shopping. Starting with Google Scholar, I discovered that, in literature, these kinds of deliveries are named "Crowd-Shipping.” Crowd-shipping is a type of delivery where non-professional drivers or commuters are engaged in delivery activities. A crowd-shipping system can be based on non-professional drivers or a mixture of crowd and professional drivers.
Compared with standard delivery systems, crowd-shipping is a cheap, fast, and sustainable type of delivery. It is an innovative socio-economical last-mile delivery concept. Its association with sharing economies has made it very popular [1]. However, each new system has its challenges.
In a crowd-shipping system, the drivers are not committed to doing their delivery tasks. So, the system might not seem reliable [1]. Although different platforms try to deal with this, the unreliability rate is not negligible [2]. Therefore, using professional drivers as a backup plan is needed to maximize responsiveness. In addition, the unreliability of occasional couriers (non-professional drivers) brings uncertainty. Operations research tools help manage these systems. Optimization techniques, such as two-stage stochastic optimization, can significantly support decision-makers in dealing with challenges like uncertainty.
Some researchers have used two-stage stochastic optimization to solve different types of crowd-shipping problems. The work by [4] is an excellent example of this topic. Their work is one of the most comprehensive ones. They proposed a crowd shipping platform where a retailer sells products of various sizes from a central depot. Some of these products are time-sensitive, some require a delivery signature, and some might fail to do the delivery. They formulated the problem as a vehicle routing problem with a stochastic supply of crowded vehicles and time windows. They also developed a two-stage stochastic set-partitioning formulation. The first stage was to determine the customers that have been assigned the occasional couriers. The second stage reveals the rejected delivery tasks to determine the required professional drivers’ routes.
The model then developed to dynamically assign routes to vehicles. They also calculated an upper bound for the maximum number of crowd vehicles that help predict courier drivers (CD) and propose a branch and price algorithm and column generation heuristic to rapidly provide feasible solutions for large instances of the problem. Their results showed that the obtained value of using the proposed model is 21% of the total delivery cost. Therefore, using crowd-shippers will result in significant cost reduction. Also, there is a trade-off between CD’s compensation and engagement. However, they showed that low CD’s compensation may result in up to 28% of cost reduction. They also showed that by considering CD’s rational behavior, we could let them choose their route rather than assigning them. However, irrationality may lead to 4% cost increase.
Solving a crowd-shipping problem is complex. While the shipper wants to minimize his expenses, he also needs to increase the compensation of crowd drivers to increase their willingness to do delivery tasks and diminish uncertainty. Furthermore, customers look for a cheap, fast, and reliable system. Managing these three objectives in a dynamic environment is challenging and developing fast algorithms to solve real-world instances is necessary.
Finally, crowd-shipping is a new transportation concept requiring extensive quantitative work to plan and evaluate its viability in different scenarios. Using crowd vehicles entails various sources of uncertainty that affect distribution planning. For example, the transportation capacity that non-professional drivers can bring to the last-mile delivery system may vary daily. Moreover, the probability of a driver accepting or rejecting a route depends on the other delivery offers.
This article reviews the application of crowd shipping and provides an example of how stochastic programming helps solve these problems. Since very few studies have investigated the stochastic variants of crowd-shipping, there is much room for research. Most studies assume that the compensation amount (fixed or variable) is an input parameter of the problem (see for example [3]). Hence, future studies can consider deciding on the optimal compensation level, which makes the problem of the domain of revenue management. Moreover, most studies consider compensation as a criterion for accepting or rejecting a route (see for example [1] ). Exploring other criteria, such as deviation from the actual driver route, provides an area for future research.
While I am writing this article, my package has been delivered to my loved one and a courier driver is ringing the bell to deliver my groceries. All thanks to cheap and sustainable crowd-shipping services!
References
[1] Gdowska, K., Viana, A., Pedroso, J.P., 2018. Stochastic last-mile delivery with crowdshipping. Transportation research procedia 30, 90–100.
[2] Mousavi, K., Soldouz, S.A., Roorda, M.J., 2019. Crowd-Shipping: Assessing Alternative Operational Models using the Vehicle Routing Problem with Occasional Drivers. Technical Report.
[3] Torres, F., Gendreau, M., Rei, W., 2022a. Crowdshipping: An open vrp variant with stochastic destinations. Transportation Research Part C: Emerging Technologies 140, 103677.
[4] Torres, F., Gendreau, M., Rei, W., 2022b.