Project

Autonomous micro-mobility study for food delivery

Groups

Could we substitute cars by shared autonomous lightweight vehicles for food delivery? We have developed an agent-based simulation model that analyzes the performance of  shared autonomous micro-mobility compared to the traditional car-based deliveries.

The population growth and the technology development are leading to new realities. One of them is the increase of online food deliveries, which are usually done by traditional combustion cars in the US. In the City Science research group we are exploring new vehicles to accomplish this task.

Could we substitute cars by shared autonomous lightweight vehicles for food delivery? We have developed an agent-based simulation model that analyzes the performance of  shared autonomous micro-mobility compared to the traditional car-based deliveries.

The population growth and the technology development are leading to new realities. One of them is the increase of online food deliveries, which are usually done by traditional combustion cars in the US. In the City Science research group we are exploring new vehicles to accomplish this task.

The MIT City Science has already developed on-demand autonomous micro-mobility vehicles such as the PEV and the MIT Autonomous Bicycle. In a previous study we saw that even if autonomy can increase vehicle usage times from below 2% to above 8% there is still a large percentage of the time that vehicles are not used. For this reason, we realized there is potential to envision new uses for these vehicles during off-peak hours.

In this study we are exploring the possibility of utilizing these vehicles for food deliveries, which are traditionally done by cars in the US. Substituting traditional combustion cars for lightweight autonomous electric vehicles could potentially bring benefits in terms of CO2 emissions. To achieve our goal, we have built a multi-layer agent-based simulation model in GAMA platform. This model can be used to size each vehicle’s fleet needed to answer the food delivery demand previously generated, to analyze the average distance traveled by each system, or to assess the corresponding environmental impacts, among others. Moreover, the model allows to introduce different types of vehicles, vary autonomous driving speeds, battery lives, and recharging techniques.

The main results regarding the fleet size of the different food delivery systems needed in order to answer the same demand with the same quality standard (>95% of the trips on time), are shown in the following diagrams.

Results show a significant dependency on the charging techniques used for the shared autonomous micro-mobility systems when sizing. Furthermore, the autonomous vehicles' speed rate and battery life also influence in the minimum fleet needed to answer the defined quality standard. Increasing the riding speed results in smaller fleet sizes needed with any of the two charging technologies (swapping and conventional charge). Increasing the battery life, instead, clearly influences the sizing of conventional charge vehicles but does not affect the systems with swapping technology.

In this research, we have also run environmental impact analyses of the different systems. Main results are gathered in the following diagram.

Future steps of this study include, among others:

  1. Detailed life cycle assessment of the proposed systems.
  2. Infrastructure sizing.
  3. Expansion to global package delivery.
  4. Study of the performance of these vehicles when used for movement of people and food deliveries at the same time.