Research

Lead Researcher

Gregory Stevens
President, Causality LLC; Research Director, Mcity

Project Team

Karl Wunderlich
Director of the Surface Transportation Systems Division, Noblis Autonomous Systems Research Center

Project Abstract

Depot operations for fleet vehicles include a range of sequential activities, including vehicle loading and unloading, charging/refueling, maintenance/safety checking, and marshalling/dispatching. These operations are conducted sequentially throughout the workday but are subject to peak demand periods where valuable driver/operator and vehicle time is wasted queueing at activity stations. Further, queued vehicles obstruct the movement of depot staff and vehicles-in-motion, resulting in both negative safety and efficiency conditions. Under exceptional demand conditions, queued vehicles can produce gridlock conditions within the facility, dramatically reducing depot productivity. In this research activity, Noblis and colleagues at the University of Michigan (U-M) propose to automate within-the-fence line low-speed depot vehicle motion among activity stations using a combination of existing vehicle SAE Level 2 automation/ ADAS and relatively low-cost infrastructure sensor systems deployed throughout the depot yard. Automated vehicles are expected to navigate safely at low speed between stations/ingress/egress points, avoiding collisions with depot infrastructure, other vehicles, and (suitably trained) human depot staff in motion within the facility.

The hypothesis for this work is that an adapted form of the existing Noblis Orchestrated Autonomy (NOA) code base can be utilized as a form of overall strategic control in the system superior to centralized forms of control in terms of safety and depot productivity, namely:

• Priority and multi-machine path planning are determined using decentralized NOA bidding/assignment module
• Joint maneuver plans can be established based on urgency/time criticality
• The depot is likely to have multiple heterogenous machines in the system in various states of good repair, so monitoring maneuver and sensor performance over time could be a safety-related attribute in this environment
• Humans working in the space with automated vehicles/machines can also be represented/integrated intelligently and not just as “dark objects” to sense and avoid

U-M has developed a virtual Automated Depot prototype which will be the test-bed for the FY25 activity, and in which the NOA code base can be interfaced. U-M plans a physical prototype facility to be ready in the FY26 time frame for potential follow-on development/application.

Outcomes: Overall, this research project will strengthen Noblis engagement with academic partners, provide a compelling test case for the benefits of NOA, and has general value in application for depot/warehouse automation for current Noblis clients (DoD, USDOT).

Project Outcome

The hypothesis for this work is that an adapted form of the existing Noblis Orchestrated Autonomy (NOA) code base can be utilized as a form of overall strategic control in the system superior to centralized forms of control in terms of safety and depot productivity. This research project will strengthen Noblis engagement with academic partners, provide a compelling test case for the benefits of NOA, and has general value in application for depot/warehouse automation for current Noblis clients (DoD, USDOT).