Building a Robot-Friendly Environment: Insights from Research on Infrastructure for Robot Delivery

Building a Robot-Friendly Environment: Insights from Research on Infrastructure for Robot Delivery
Img. credit: Northern Arizona University

College campuses are bustling with various modes of transportation, from walkers and runners to cyclists and skateboarders. These days, one can also spot six-wheeled robots zipping around, offering contactless food delivery. However, just like any busy sidewalk, these robots introduce the potential for collisions and unsafe interactions with pedestrians. To address these concerns, researchers from Northern Arizona University conducted a study to analyze the frequency and causes of conflicts between sidewalk autonomous delivery robots (SADRs) and pedestrians or bicyclists, with the aim of finding solutions for the future.

The study, published in Transportation Research Interdisciplinary Perspectives in March, was co-authored by Steven Gehrke (an assistant professor inthe Department of Geography, Planning and Recreation); Christopher Phair, a master's student in applied geospatial sciences; Brendan Russo, an associate professor inthe Department of Civil Engineering, Construction Management, and Environmental Engineering; and Edward Smaglik, a professor in the same department. It is the first of its kind to examine the impact of SADRs on pedestrian and bicyclist safety.

Lead author Gehrke emphasizes the need for careful decision-making when it comes to routing these robots on college campuses. By informing future planning decisions and evolving facility management practices, campuses can ensure the safe coexistence of robots and pedestrians in shared-use environments. Given the growing popularity of delivery services on campuses, this research can also serve as a guide for other institutions considering the introduction of similar services.

The researchers conducted their observations on Northern Arizona University's campus, which has been using SADRs since 2019. They set up cameras at ten sites and recorded interactions over a one-week period. The study focused on identifying human-robot conflicts, determining the initiator of the conflict, and the evasive actions taken. Interactions were classified as moderate, dangerous, or not conflicts, with dangerous interactions defined as instances where the robot and traveler crossed a shared point of the pathway in under 1.5 seconds. The researchers recorded 12 instances where the elapsed time was zero seconds, resulting in collisions.

The study found that the most serious conflicts occurred when SADRs crossed paths with pedestrians on sidewalks, either by crossing in front of them or overtaking them. The scientists also identified other factors that could impact safety, such as sidewalk width, the number of intersections, and the presence of robots, pedestrians, and cyclists in the area. Sites with narrower sidewalks and more intersections experienced a higher number of dangerous interactions.

The research suggests the importance of establishing SADR routes that prioritize parallel travel along wider sidewalks while minimizing high-activity sidewalk crossings. Designating less-trafficked, well-marked sites for robot drop-offs can also contribute to reducing interactions. Cities considering the implementation of SADRs must focus on infrastructure development, education, and route programming to minimize potential conflicts and ensure the safety of pedestrians and cyclists.

Gehrke emphasizes the significance of real-world research to provide evidence supporting programs and policies that facilitate the safe operation of SADRs in environments shared with vulnerable road users. Furthermore, more research is needed, including assisting cities in determining the appropriate fleet size to meet delivery needs without compromising pedestrian and cyclist safety in downtown street networks.

Src: Northern Arizona University