Intelligent Autonomous Ground Mobility

Recognizing the need for autonomous ground systems to operate in the unknowns of a mission, the U.S. Army is making investments in ground vehicle mobility modeling and simulation (M&S) to improve and prepare for future off-road operations. Army engineers and scientists are working diligently and purposefully to shape future operational capabilities and, as a ground force, remain ready and resilient.

By Dr. David Gorsich, CCDC GVSC and Michael Letherwood, Alion Science & Technology

From Armor & Mobility, March/April 2020

As the Army looks to the future, there is an opportunity for the ground vehicle community to help shape the unique role of land forces in achieving national security objectives. As intelligence, surveillance, target acquisition, and reconnaissance capabilities are rapidly developing, assured mobility becomes even more important. In 2016, Chief of Staff of the United States Army Gen. Mark A. Milley stated that “On the future battlefield, if you stay in one place longer than two or three hours, you will be dead. With enemy drones and sensors continuously on the hunt for targets, there won’t even be time for four hours of unbroken sleep.” The Army’s future force must be able and ready to be called upon for a variety of missions so it must be ready to apply land power/ground forces toward achieving strategic outcomes across the full range of military operations. Development and deployment of autonomous weapons systems generally point to several military advantages such as acting as a force multiplier, and, more importantly, may require fewer warfighters for a given mission. As depicted in figure 1, when vehicles become immobilized, troops are put at risk and the mission is jeopardized.

Unlike commercial autonomous systems, the military must operate in unknown and unstructured environments where roads may not exist but the supplies must reach the front lines. On the battlefield, mobility is the key to survivability and it’s crucial for commanders to know which vehicle to deploy on what terrain. Commanders need to have the ability to assess their own and opposing forces vehicle mobility in the area of operations, which will increase confidence in mission planning and reduce the risk of mission failures due to compromised vehicles. The U.S. Army’s Combat Capabilities Development Command (CCDC) Ground Vehicle Systems Center (GVSC) strategy provides an overarching framework to develop, integrate and sustain advanced manned and autonomy-enabled ground system capabilities for the current and future force.

Advancing Autonomy

For ground autonomous mobility, GVSC has shown the potential superior mobility of autonomous ground vehicles over tele-operated vehicles and is implementing a roadmap towards assessing autonomous mobility through M&S. GVSC is developing an understanding of how to leverage autonomy and autonomous systems – understanding not only the technological value of these new capabilities, but also how the off-road mobility has a huge impact on successful autonomous operation and mission completion. Autonomous systems and autonomy-enabled manned ground platforms are enabling capabilities that provide force multiplication to warfighting functions.

Mobility is regarded as a vital component of autonomy. These capabilities are major objectives of GVSC’s research and development programs as it continues to collaborate with its partners to integrate technologies and develop advanced capabilities that improve warfighter effectiveness and efficiency. The emergence of intelligent ground vehicles and their dependence upon quantitative analysis of mobility has infused terrain vehicle systems M&S with a new relevance and broader scope than ever before. Mobility metrics and analysis for robotics and vehicle intelligence (VI) is a very active and prolific research area and is an essential element of M&S from two application perspectives: 1) inclusion of robotics and VI in mobility metrics and assessments for operational planning, acquisition, and design; 2) embedding M&S models and metrics into robots and VI algorithms because they are standards for mobility assessment and decision making.

Evolving Analytical M&S Mobility Capabilities

The objective is to generate models and data products for predicting vehicle performance that can be used to plan and execute desired mission scenarios over specified regions. Beyond operational use, these capabilities can be used for autonomous vehicle development as well as the acquisition process. The current mobility assessment methodology is called the NATO Reference Mobility Model (NRMM) and is a simulation tool aimed at predicting the capability of a vehicle to move over specified terrains. It’s empirically based and developed using decades-old data and technology, but it is also broadly understood to be theoretically limited and difficult to adapt to contemporary vehicle design technologies and to implement within modern vehicle dynamic simulations. “The current model is outdated, it’s old, but it’s still useful for current systems that are different nowadays” according to GVSC’s Chief Scientist, Dr. David Gorsich. “Our vehicles have stability control systems, tire pressure systems, etc. and all those systems make a difference in your off-road mobility, and we really do need to update M&S capabilities and standards to be able to predict those types of things. Also, high performance computers and simulation have come a long way since the 70s that allows the implementation of complex soil mechanics where the physics of the soil and the interaction of vehicle and soil are being considered. There’s a lot that can be done if we understand the variability in the soil and the terrain when predicting mobility.”

Dr. Gorsich further added, “The NATO reference mobility model is so important, from an acquisition perspective, that when we buy systems we need to understand how good they are from a mobility perspective. But also in operation, we need to understand how well a vehicle can go from point A to point B and can it carry out this operation with this vehicle or a set of vehicles.” To address the problem, GVSC researchers and engineers partnered with a NATO Research Task Group (RTG), which consisted of 70 members from 15 nations, to develop a Next Generation NATO Reference Mobility Model (NG-NRMM). NG-NRMM is defined to be any M&S capability that predicts land and amphibious vehicle mobility through coordinated interoperation of Geographic Information Systems (GIS) software and multibody, physics-based vehicle dynamics M&S software. NG-NRMM is a new capability that lacks extensive experience and maturity and its’ development involves rapidly evolving technologies and scope. The physics of vehicle-terrain interaction is better understood today due to the advancement of M&S capabilities. As depicted in figure 4, the goal is to place the physics-based mobility software at the center of the geospatial terrain data and soil maps so that mobility performance metrics such as a Go/No-Go map (which will be explained later) can be derived. This mobility metric can be used in the acquisition process and in operational planning as is done today using NRMM.

The M&S software must be capable of utilizing terramechanics to properly assess vehicle-terrain soft soil interactions, incorporate capabilities to portray autonomous control systems, and include uncertainty quantification (UQ) to enable probabilistic M&S. Terramechanics modelling is focused on vehicle terrain interaction that accounts for soft soil (i.e. deformable soil) effects on mobility. NG-NRMM has the potential to significantly reduce procurement risks enabling alternative solutions to be considered and will also provide operational decision makers with a tool for assessing their own and opposing vehicle mobility in the area of operations, which will increase confidence in mission planning and reduce the risk of mission failures due to compromised vehicles. “This research could prove relevant when it comes to the change in the character of warfare, the kinds of operations that NATO member states have been part of, as well as due to technology developments,” said Marta Kepe, a Rand think tank analyst specializing in defense, security, and infrastructure. A myriad of other complexities complicate the movement of military land platforms across Europe, including “the conditions of transport infrastructure; multinational, cross-border and national level-movement coordination, including between military users of infrastructure and the civilian managers; and national legal requirements,” Kepe added. The vision is to reach a point where nearly all virtual prototyping and operational effectiveness can be determined up front leading to rapid fielding of technology with a clear understanding of the operational capability of the technology. NATO Secretary General Jens Stoltenberg, in speaking before the alliance’s Parliamentary Assembly on May 28, 2019 said NATO’s security “does not just depend on the forces we have deployed, but it also very much depends on our ability to move forces to reinforce quickly if needed.” The goal of M&S investments is to minimize the need to build physical prototypes, and to fill the gaps in our mobility M&S capabilities especially for autonomous operations.

NG-NRMM Software

For effective autonomous navigation, NG-NRMM software tools must be capable of predicting a real vehicle’s mobility results on any given terrain map to support operational analysis and mission planning purposes, to include selecting the optimum vehicle path on a terrain map based on the mission requirements. It must also be capable of replicating the existing NRMM output products which includes: Go/NoGo trafficability and Speed-Made-Good (SPG) maps as well as speed limiting reason codes and single-pass/multi-pass results. The output results are in two categories – Go areas and NoGo areas. Go/NoGo maps identify areas where the modelled vehicle can and cannot go. The Go areas are usually portrayed as “green” areas on the map, while NoGo areas are normally portrayed as either “red” or “black”.

NG-NRMM also must generate a list of “reason codes” that provide further insight into the causes behind a vehicle’s immobilization. These additional insights can shape route planning, choice of a vehicle for a selected mission, and inform vehicle acquisition / modernization decisions. Example reason codes include: inability to negotiate / overcome obstacles; inability to negotiate vegetation; and inability to overcome soft soil / slope resistances. Lastly, NG-NRMM must be capable of predicting maximum safe speed for each terrain unit. SPG maps enable users to quickly and easily determine the best areas to conduct operations. Other newly desired output metric capabilities also included generating results for vehicle stability/handling, urban manoeuvrability, path modeling, fuel consumption/range estimation, and rut depths.

NG-NRMM is intended to expand the basis of the legacy NRMM to define innovative M&S mobility capability that develops, such as autonomous navigation, and facilitates interoperation with current and evolving M&S capabilities. Figure 6 depicts the flow of data through the NG-NRMM analysis process.

First, GIS data is collected and aggregated into a file geodatabase using standard GIS tools and processes. In order to achieve this, the NG-NRMM imports and aggregates remotely-sensed GIS data and generates terrains that can be analyzed in the NG-NRMM vehicle / terramechanical analysis software. The data in the file geodatabase are processed to generate the terrain properties needed by the multibody, physics-based vehicle dynamic M&S software. Capturing the accurate soil mechanical properties such as internal friction and cohesion are critical to evaluating soft soils and for vehicle terrain interaction and this is possible with physics based terramechanics modeling. The multibody dynamic vehicle M&S software executes vehicle runs using the terrain files and generates results for each terrain unit. NG-NRMM compliant software preserves the spatial orientation of the data by linking the results to the original terrain file. Using GIS software, the data can now be visualized to produce spatially-oriented, map products. GIS data is critical to building the required terrains needed to support coalition mission planning and operational effectiveness analyses required for autonomous operations.

Once the prototype software was developed, the RTG conducted a virtual demonstration which was an “end to end software demo” that demonstrated how NG-NRMM adopted new technologies, modelling techniques, and computational tools to enable physics-based simulation of any vehicle design, in complex environments and scenarios. The team also conducted a set of verification and validation (V&V) field exercises, using both a tracked and wheeled vehicles, to evaluate the state of the art terramechanical models.

LOOKING AHEAD

Even as full autonomy remains the eventual goal, essential to the reliable operation of autonomous vehicles in the field to successfully carry out a mission is the ability to predict its mobility performance and risk over a specified region. Such predictive capability is needed to effectively monitor and guide autonomous vehicles to keep the vehicle safe while meeting mission constraints (e.g., no-go areas) and maximizing performance metrics (e.g., time, speed, fuel consumption). The viable use of autonomous vehicles depends on the development of predictive models and data products that can guide the vehicle safely and effectively in the field. The future of analytical soft soil mobility analysis clearly rests with NG-NRMM as it holds the promise of allowing manufacturers, planners, and users the ability to model virtually any platform, over any soil and terrain type. NG-NRMM is vital to the Army’s mission as it will add new capabilities in the design, modeling, and simulation of a broad class of vehicles, with the potential to reduce costs and improve performance. Future intelligent autonomous mobility may involve many different classes and sizes of vehicles such as wheeled/tracked vehicles, small robots, legged robots, humanoid robots, and other emerging technologies traversing a variety of environments that may include on-road, urban, off-road, and building interiors. NG-NRMM could yield a new paradigm for ground vehicle mobility with the possibility to model complex vehicle maneuvers in high fidelity. The mobility performance metric maps generated using this technology are key requisites for consideration of military missions which could succeed or fail depending on how accurately the performance maps are generated.