Evaluating the Impacts of Autonomous Delivery Robots (ADRs) on Pedestrians and Vehicle Traffic

Principal Investigator

Miguel Jaller

Description

Autonomous robots have been operating inside warehouses and factories (controlled environments) for more than three decades, demonstrating significant cost and time savings. Motivated by these benefits and potential improvements in sustainability, developers, couriers and retailers have recently started developing and testing robots for last-mile deliveries. However, the dynamics of the urban environment represent new challenges for the operations of these robots. For instance, the operation and the limitations of autonomous delivery robots (ADRs) (e.g., signal loses, acceleration, reaction time) could disrupt pedestrian and vehicle traffic. Specifically, such disruptions could include the interaction and potential conflicts of ADRs with vulnerable populations such as disabled individuals, elderly, and children, when sharing the sidewalk. Moreover, when ADRs cross an intersection, they have a similar or higher probability of being struck by a vehicle, getting stuck in the pedestrian flow, and then causing delays in the traffic crossing the intersection. These delays could be analyzed as an extension of the red-light time, which definitely represents a major mobility and safety disruption, especially during rush hours. A lack of understanding about these issues hampers the development of policies and strategies to enable safe operation of ADRs. More  importantly, the lack of understanding creates uncertainties about their role in a sustainable transportation system. This project has two main goals: (i) to improve understanding of the impacts of ADRs operation on sidewalks and intersections, and (ii) to lay down the key features,—required technology improvements and changes in the urban infrastructure—to operate ADRs in a non-disruptive way for vehicle traffic and pedestrians. The results will inform ADR development as part of the new mobility revolutions, as well as, provide insights to policymakers to design plans, policies, or initiatives for safe and efficient accommodation of this technology. The team conducted an exploratory study with real ADRs belonging to the startup KiwiCampus to understand the robots’ features. This provided a preliminary understanding of the interaction between ADRs and pedestrians and highlighted the potential conflicts with vehicle traffic. Using measured parameters such as speed, acceleration/deceleration rate, and recognition patterns, the team developed a simple simulation model to evaluate such interactions. The simulations showed that the ADRs could negatively impact urban mobility (on- and off-street). For instance, in 43% of 1000 random trials, the ADR was delayed crossing an intersection, thus blocking it after the light changed to green, generating significant queues of vehicles. Among these trials, 5% of the delayed instances, showed delays greater than 10 seconds, 10% between 5 and 10 seconds, and 43% between 1 and 5 seconds. Although, the model considered a number of assumptions, the empirical results were consistent with the anecdotal observations from KiwiCampus operations in Berkeley, CA. In this project we will conduct an in-depth analysis of these issues, by evaluating different operational scenarios with KiwiCampus ADRs, to determine the key variables to mitigate the risk of conflicts on sidewalks and intersections. Specifically, we aim to determine strategies that allow reducing the safe crossing time of intersections and mitigating conflicts with pedestrians (other users). In doing so, we will conduct real experiments and improve the simulation tool.

Acknowledgements

 

Resources