Grant: Discovery Grants Program - Individual

The global same-day delivery market has witnessed significant expansion, with its valuation expected to grow to $26.4 billion USD by 2027, more than triple its 2021 baseline. While this surging demand offers opportunities, it also presents significant operational challenges. In recent years, Crowdsourced Delivery (CSD) has emerged as an innovative last-mile delivery method, where freelance independent drivers complete delivery tasks using their own vehicles. Despite its potential benefits of achieving low-cost delivery leveraging existing vehicles, the self-scheduling nature of drivers introduces additional uncertainty, compounding the existing challenges of traditional truck-based delivery systems. This research program aims to bridge gaps in CSD literature by examining network design decisions, considering worker welfare and matching equity, integrating sustainability factors, and employing diverse modeling techniques to capture uncertainty. The research will investigate three main themes: (1) Designing efficient crowdsourced delivery systems. This involves examining optimal network designs that capitalize on existing traffic flows and integrate environmentally-friendly delivery means. We will develop robust and stochastic optimization models, that consider uncertain driver availability to determine where service should be offered and to design multi-tier networks that enable the use of greener delivery methods. (2) Optimizing crowdsourced delivery operations. This theme delves into the operational intricacies of CSD, considering driver matching equity, scheduling a subset of drivers while accounting for drivers' spatial distribution, and balancing profitability, service reliability, and driver welfare. (3) Advances in optimization under uncertainty motivated by CSD problem settings. This theme will investigate different ways of modeling uncertainty addressing CSD problems. It will explore the relationship between platform decisions and uncertain parameters like driver and order arrivals, by modeling endogenous uncertainty. It will also examine integrating prediction models within optimization frameworks to anticipate the most likely uncertain outcomes. This research program will make significant progress in designing the technology needed for crafting efficient, reliable and sustainable last-mile delivery systems that leverage crowdsourcing, potentially elevating delivery solutions not just in Canada but globally. Moreover, the developed models will offer valuable insights and recommendations to both practitioners and policymakers, promoting informed and socially-responsible management of crowdsourced delivery platforms. Finally, the program will advance the domain of modeling uncertainty in optimization problems, by considering different types of uncertainty and integrating predictive and prescriptive analytics, which will have broader contributions and applicability that extend beyond the transportation domain.