Use of MnCORS to Support AV Operations in Rural Minnesota

Principal Investigator(s):

Rajesh Rajamani, Professor, Chair, Mechanical Engineering

Co-Investigators:

  • Brian Davis, Associate Dir., Mobility Tech. Lab, Mechanical Engineering

Project summary:

The Minnesota Continually Operating Reference Station (MnCORS) network is a service operated by the Minnesota Department of Transportation that consists of a network of base stations across the state which provide global navigation satellite system (GNSS) correction signals to real-time kinematic (RTK)-capable GNSS receivers. The objectives of this project are to use localization measurements from the MnCORS GNSS system to demonstrate key autonomous vehicle (AV) operations on all types of roads in rural Minnesota (highways, unmarked paved roads, and gravel roads). Further, the project will demonstrate and characterize the limitations of localization using this network when it comes to AV autonomous operations in urban locations.

Autonomous vehicles (AVs) typically perform steering control using lateral distance to lane markers measured by cameras. Since low-volume rural roads may not have lane lines, may not even have center lines, may consist of gravel or unpaved portions, and may not be plowed for snow removal, AVs have great challenges in operating on such roads. Further, rural intersections may be missing signs that are critical for AV navigation and control. This project will conduct an experimental study on localization using the MnCORS GNSS network to compensate for the lack of lane markers and intersection signs on all types of rural roads. The use of GNSS is motivated by the fact that rural locations have open skies and low-multipath environments that can provide reliable connectivity for RTK-capable GNSS receivers. MnCORS corrections and sensor fusion with low-cost inertial sensors will be used to obtain cm-level accuracy for steering control.

A related previous project at the University of Minnesota has developed a snowplow driver assistance system in which GNSS-based position measurement is used to display the snowplow's lateral position inside the lane. This is being used to provide guidance to the driver in poor visibility conditions and has been very popular with MnDOT snowplow operators. The lateral position guidance to the operator is provided in one-foot increments, and its usage is restricted to times when an RTK fixed-integer solution is achieved, which is a function of cellular access to the MnCORS network. This project plans to build on the previous project, demonstrate the capability to achieve continuous cm-level accuracy using a combination of MnCORS GNSS and inertial sensor measurements, and utilize this estimated position for reliable automatic steering control. The technology will also provide accurate localization with respect to intersections and enable autonomous operation at intersections, including turns.

The project will conduct a pilot study that demonstrates automatic steering control at two locations using the MnCORS GNSS system on the MnCAV autonomous vehicle. The locations for the demonstration will be chosen in consultation with the project's technical assistance program (TAP). The demonstration will show automatic steering control on rural paved and unpaved roads lacking lane boundaries. Additionally, the project will also conduct a winter driving demonstration using the same technology to show driving on snow-covered roads when lane boundaries are not visible.

In addition to demonstrating autonomous mode operation in rural locations, the project will also characterize the overall capabilities and limitations of localization using the MnCORS network and the RTK-capable GNSS receivers on the MnCAV vehicle. This characterization will include metrics of accuracy, real-time update rates, convergence times, network availability, and sources of failure both in rural and various types of urban locations. The urban/suburban locations taken up for characterization will include downtown Minneapolis, downtown Saint Paul, urban locations without tall buildings, suburbs, and underneath overpasses/tunnels.

Project details: