Agile Ground Vehicle Dynamics, Energy Efficiency, and Performance in Severe Environments

International Symposium

Hilton Birmingham Perimeter Park Hotel
Birmingham, AL September 8 - 11, 2013

SESSION 1: Road and Terrain Ground Vehicle Dynamics: Past, Present, and Future Trends
Session Chair: Vladimir Vantsevich, University of Alabama at Birmingham

Following the introduction of self-propelled motor vehicles over a century ago, the increase in travel speed confronted engineers with the need to better understand the dynamics of wheeled vehicles. Therein began the development of engineering methods to design powertrains to propel the vehicles, brake systems to stop them, suspension systems to allow negotiation of imperfect roads, and steering systems to control their direction. Characteristic of all the on-road vehicles of interest here is that they are carried on rubber tires – one of the most important components affecting their agility.

Agility, a term that will be heard often at this symposium, might be defined here as the ability to move in a quick and controlled fashion. Since tires are the principal source of forces to control vehicle movement, characterizing tire properties has been an essential starting point for understanding vehicle dynamics and performance. Hence one of the goals in this presentation will be to define the tire properties most critical to each mode of performance.

On-road vehicle dynamics are generally classified as acceleration, braking, ride, handling and rollover. In the context of this symposium, acceleration also encompasses mobility on difficult terrains, startability, and gradeability (hill climbing). In a similar fashion, braking involves not only stopping the vehicle, but negotiation of downhill grades. Handling, which involves the ability to turn the vehicle, is also closely linked to maneuverability – i.e., the ability to negotiate tight quarters at low speeds.

Relatively simple engineering models of the mechanics associated with each mode can be used to quantify quasi-static limits of performance by equations. The beauty of representation by equations is that in the eyes of the analyst they identify the principal vehicle properties influencing performance and the way in which they affect performance. When more precise and complex modeling is required, particularly to deal with transient behavior and non-linear systems, simulation of the vehicle dynamic behavior has become the tool of choice. Examples will be given.

In addition to analysis tools, the practitioner needs to be aware of the tests that can be used to evaluate each mode of performance and what metrics are used to quantify the result. There are two major classes of performance tests – a general menu of tests that are used by the automotive community to evaluate dynamic performance and specific tests that are required by various government regulatory agencies to ensure safety of on-road vehicles. The general menu is quite extensive so only a sampling of the more common tests will be described. The primary government tests will also be described.

The presentation ends with a glimpse into the future, looking at the next generation of dynamic controls for on-road vehicles.

There is a virtual revolution underway for vehicle development with regard to efficiency and emissions being driven by U.S. government policy and the force of the U.S. and European market place. The real dilemma is that the cost of these vehicles continues to increase just as computer costs increased in the decades of 1960-80. By analogy, it is time to open up the architecture of vehicles, standardize all principal components (power source, electrical generation, batteries, super caps, in-wheel drives, domain-specific operational software, etc.) in a minimum set, invite multiple suppliers to participate in a responsive supply chain, encourage third party components for after-market vehicle refreshment, etc. to drive down cost while continuing to improve vehicle efficiency and safety.

Contrary to recent examples (La Ferrari, Tesla-Sedan, Via Motors, etc.), performance does not necessarily require higher cost. If that were so, social media would not exist. Two principal developments are required:

1.         A full open architecture that combines both power generation and power utilization.
2.         A data dense interface between the vehicle operator and the intelligent vehicle.

In the first development, almost all present activity is on power generation leaving utilization virtually untouched. Utilization can equally improve efficiency but it can also dramatically improve traction to enhance safety for poor surface conditions, emergency maneuvers, and off-road operations. In the second development, a fully instrumented driver and intelligent vehicle can combine their documented performance capabilities to provide for autonomy (if desired), an auto-pilot mode, or a fully immersed operator to maximize responsiveness to emergencies.

It is now possible to create a strategy for long-term vehicle development to enhance performance while driving down cost, getting young engineers and scientists excited, and guide national investments to most effectively move the tech base forward (without tangential programs which distract the community from useful development activity) that is based on real requirements in the market place.

As was proven in analytical and experimental research, the dynamics, energy efficiency, and performance of multi-wheel drive vehicles with four or more wheels depends on power distribution between the drive wheels. By controlling the driveline system, which splits power between the drive wheels, one can control vehicle dynamics, fuel/energy consumption, and vehicle performance. This presentation illustrates the above-said by case studies and also provides analytical material that can be recommended for modeling and simulation of vehicles and for designing driveline systems.

The presentation is specifically concentrated on the interaction between the driveline and other vehicle systems involved in the wheel power distribution process, such as the brake system, suspension, and steering system. It is shown that the systems can both contribute and interfere with each other when influencing the wheel power distribution and impacting vehicle dynamics, performance, and energy efficiency. This interaction/interfering of the systems becomes crucial in severe environment conditions when a control decision should be made and implemented in milliseconds. Thus, such features of agility, such as pre-emptive, fast, and exact interactions between the vehicle and environment, make it an important subject to be researched.

The presentation explains the fusion of the vehicle systems’ actions when a vehicle is in motion. A passive functional/operational fusion of the driveline-brake-suspension combination is considered in detail. A novel approach, an active operational fusion, is presented to improve both energy efficiency and lateral dynamics of an all-wheel drive vehicle by means of active functional/operational fusion of a driveline system and a steering system that steers the front driving wheels.


SESSION 2: Modeling of Multi-Physics, Fast/Slow, Interdependent/Independent Systems
Session Chair: Azim Eskandarian, George Washington University

In order to develop sophisticated controls for vehicle dynamics, we first need reasonable low order, conceptual models that allow inclusion of actuators along with the representative vehicle dynamics. The actuators might include electromechanical or electro-hydraulic devices in the suspension locations, electric motors for power and braking, electro-mechanical steering actuators, controlled differentials, and more. The challenge is to include the actuator dynamics along with the vehicle dynamics in a single model so that integrated control concepts can be developed and ultimately tested with hardware-in-the-loop development tools.

The modeling tool that best enables the modeler to include all energy domains in a single overall model is bond graphs. In the presentation, bond graph modeling will be briefly introduced and their use will be demonstrated for vehicle models of increasing sophistication. Some examples of their use in controller development will be presented.

Multi-agent systems arise in several domains of engineering and they can be used to solve problems which are difficult for an individual agent to solve. Examples of multi-agent systems include disaster response, online trading, and modeling social structures. For the military, heterogeneous networked teams need to perform in a reliable manner under changing mission requirements, varying resource reliability, and resource faults. Strategies for team decision problems, including optimal control, N-player games (H-infinity control, non-zero sum), and so on, are normally solved for off-line by solving associated matrix equations such as the coupled Riccati equations or coupled Hamilton-Jacobi equations. However, using that approach, players cannot change their objectives online in real time without calling for a completely new off-line solution for the new strategies. Therefore, in this presentation, we show an online gaming algorithm based on policy iteration to solve the continuous-time (CT) distributed multi-agent games for cooperative systems with infinite horizon cost with known dynamics. That is, the algorithm learns online in real-time the solution to the game design cooperative coupled HJ equations. This allows for truly dynamical team decisions where objective functions can change in real time and the system dynamics can be time-varying.

Renaissance engineers/scientists are to be trained not only on how things work but also on how the world works with creative, innovative and entrepreneur spirit. They must face societal challenges: Energy, Health and Environmental sustainability of 21st century information technology based globally competitive world. In response, several academic and research institutions have undertaken strategic initiatives to reform education and research with global collaborative alliance and with an emphasis on: IT/e-commerce, Globalization, Corporate downsizing, Rate of Social Change, Industry Standards, Multiple Thinking, Environment & Safety, Foreign Standards, Mobile work force and science and engineering innovations.

The progress realized in this endeavor of leveraging globalization, innovation and IT in Transforming vehicle technology will be presented. The emphasis will be placed on academic design projects in collaboration with industry, the tools/technology and infrastructure development essential for integration of experiments, computational simulations and theory in education, research and design, the development of virtual reality, higher dimensional visualization and associated enabling technology tools, and multidisciplinary engineering with liberal art and business and communication/soft skills blending.

The development of GI3 (Global Interdisciplinary Institute for Innovations) with global partnership involving US, S. Korea, India, Argentina and Italy will be described. The current and future directions with emphasis on global alliance for training Renaissance engineers/scientists involving industry and government partnerships for training Renaissance workforce at all levels will be included.


SESSION 3: Mechatronics Foundation for Agile Vehicle Applications
Session Chair: Jianbo Lu, Ford Motor Company

Mechatronics is a synergistic combination of precision mechanics, electronics, controls, and computer engineering, combined through a process of integrated design. A mechatronics engineer needs to be able to handle a wide range of technologies in sensors, actuators, interface hardware, control systems, analog and digital electronics, software engineering, and concurrent design. A vehicle is essentially a mechatronic system. The agility in vehicle performance comes from both dynamics and control. A satisfactory model that includes all the key domains such as mechanical, electrical-electronic, thermal, and fluid is useful in the analysis, design, control, and evaluation of the vehicle. Model optimization will be important in the realization of ability. Sensing, actuation, and associated instrumentation, and their performance and fault tolerance will be key contributors as well. Rapid and fault-tolerant control is essential in agile dynamics of a vehicle. Designing self-powered systems such as self-powered sensors, actuators, and control systems using energy harvesting techniques enables improving the efficiency and sustainability of a vehicle system. Further design improvements can be made by employing state-of-the-art techniques of biomimetics. These techniques of energy harvesting also may be treated under the umbrella of Mechatronics. This talk will introduce key issues of Mechatronics that are useful in the modeling, design and control of ground vehicles for agile performance. Particular attention will be given to the application of energy harvesting in advanced vehicle systems. Applicable concepts will be illustrated through examples.

Though attentions are focused on advanced electric and hybrid vehicles from the environmental points of view, the electric powered vehicles (EPV) have the advantages not only of low impact to the environment but also of their motion to be controlled more easily than conventional vehicles for high agility and energy efficiency. Once an EPV is equipped with four in-wheel motors for each wheel, it is easy to control the four tire longitudinal forces independently for more sophisticated vehicle motion control. In addition, the four wheels of the vehicle are easily possible to be steered independently by some electromagnetic actuators to control four tire lateral forces for the motion control. Then EPV will easily become a full drive-by-wire vehicle, which has the eight control variables (4 longitudinal forces and 4 lateral forces). It must be an ultimate individual vehicle motion control for new era.

Thus the lateral and longitudinal forces of the four wheels are controlled independently to follow basically any given vehicle response to steer and acceleration/brake commands. Though any vehicle response is available for this vehicle by appropriately controlling lateral and longitudinal forces of each tire, some reasonable tire force distribution algorithms to each tire with some norms are introduced here and the effects of the distribution controls are investigated using the experimental vehicle on a proving ground as well as with the computer simulation. The effects of the control are evaluated not only from a view point of vehicle response itself but also of reducing total tire energy dissipation due to tire slip during vehicle motion.

For the purpose of evaluating the tire energy dissipation due to tire slip during vehicle motion, the estimation method of tire slip energy dissipation is proposed. One of the advantages of the physical tire model is that horizontal tire forces are calculated at adhesion region and slip one in tire contact surface independently and also both slip speed and its direction are given by analytical formulas respectively. Based on this advantage, the algorithm to calculate the tire dissipation energy at the slip region during vehicle motion is introduced using a combined slip type of semi-empirical brush tire model.

A lateral force and a yaw moment required for the vehicle with the distribution control to follow the model response given are obtained in the control algorithm by applying an inverse method to the two degree of freedom vehicle plane motion. A total longitudinal force required for the vehicle motion is obtained from the driver’s command to braking or acceleration. The lateral force, the yaw moment and the longitudinal force obtained above must respectively be equal to the total force of the tire lateral forces, the total yaw moment produced by the tire forces and the total tire longitudinal forces. They are the 3 described constraints for the tire force distribution controls to be satisfied by distributed tire forces.

For the tire force distribution control, three types of cost functions are provided as a minimization norm of the distribution. The first one is the square sum of each tire workload which is expected to contribute to extending the vehicle performance limit and high agility with less tire energy dissipation. The second one is a simplified sum of the energy dissipation rate of each tire and the last one is the square sum of the exact tire energy dissipation rate of each tire, both of which are expected to make the tire energy dissipation as small as possible.

As the form of the function becomes the second order algebraic equation of the tire lateral and longitudinal forces which are the 8 variables in the distribution controls, the problem becomes the minimization of the second order function of 8 variables with three constraints, whichever the cost function is. Thus, partial derivative of the objective function brings us the linear first order algebraic equations of 5 variables as the minimizing condition. Solving the equations with the constraints, the lateral and longitudinal tire forces of each tire, which means the tire force to be distributed to each tire, are obtained at each moment during the vehicle motion to follow the model response. Then the traction torque and steer angle of each tire to generate the above forces obtained are determined as control commands of the distribution control.

Before the experimental validation on a proving ground, the effects of the distribution control especially on the tire energy dissipation due to tire slip during vehicle motion is evaluated with the computer simulation. In addition the effects of the distribution control are experimentally substantiated on a proving ground using the newly developed experimental full-drive-by-wire electric vehicle.

Though the distribution control with the cost function of tire workload is contributive to extend the vehicle limit performance, the distribution control does not necessarily bring us an expected effect from the tire energy dissipation view point. On the other hand, the distribution control minimizing the exact square sum of the dissipation rate shows significant effect on reducing the tire energy dissipation. Almost the same effect of the control minimizing the simplified energy dissipation rate is obtained during a moderate vehicle motion. While the vehicles without control and especially with the distribution control minimizing the tire workload dissipate the tire energy largely at the front tires, the vehicles with the controls minimizing the simplified dissipation rate and especially the exact square sum of the dissipation rate dissipate the tire energy rather equally at front and rear tires.

It is concluded that the energy efficient motion control is possible for the full drive-by-wire EPV, which will be significantly contributive to reducing the overall tire wear during vehicle motion. Also the results suggest that some extended motion control by the tire force distribution controls with some other norms and the additional constraints will be available for higher vehicle agility and active safety. For example, as a tire longitudinal force can produce the vertical force to the vehicle body through suspension mechanism, a roll motion control for handling agility is available by introducing additional constraint for the tire forces to reduce the roll motion. It will be essential to apply, in this case, the distribution norm minimizing tire work-load in order to use the four tires equally as much as possible and to avoid a specific tire being fallen into under extraordinary severe load condition during vehicle motion. Another possibility of extending the tire force distribution control is to adjust the model response for the vehicle with distribution control to follow. If the vehicle with high handling agility is preferred, a large value of stationary gain and lead time constant of yaw rate will be recommended, and then together with the distribution norm of minimizing the dissipation energy, it will bring us high agility with energy efficient vehicle. On the other hand, if the driver prefers a gentle vehicle response, relatively large value of natural frequency with appropriate damping coefficient of the model response will be reasonable. All of the above will contribute to the vehicle chassis design for fun-to-drive in a near future.

A vehicle’s agility or capability to rapidly respond to sudden changes is the key to the success of race and safe driving (e.g., accident avoidance). While race vehicles can be designed to passively achieve great agility under the control of expert drivers, normal passenger vehicles do not have this kind of “physical fitness” and their performances are subject to the limitations on the drivers’ capability. With increased implementation of chassis control systems and vehicle motion/environment sensing systems, the agility of traditional vehicles can be enhanced. In this talk, we use several examples to illustrate the effectiveness of vehicle agility enhancement through chassis controls and its application in active safety functions.


SESSION 4: Trafficability and Terramechanics in Vehicle Agility Studies
Session Chair: Corina Sandu, Virginia Tech

The current problems of off-road vehicles design will be shown in connection with improving of mobility and passing ability over the snow surface. The basic principles of the vehicles dynamic theory will be presented with the description of peculiarities of the interaction process between movers (wheeled and tracked) and non-cohesive terrains. The special attention will be paid to the problem of improvement of performance characteristics of off-road vehicles by using of intelligent systems for dynamics control. The promising areas of research related to the study of methods and algorithms for unmanned vehicles control will be considered taking in account vehicles design parameters and the environment characteristics. Such intelligent systems could ensure the best performance of the off-road vehicle during the movement over the snow surface, as well as to minimize the probability of critical situations related to the loss of the mobility and decreased passing ability. The current developments of Russian scientists in the field of computational and experimental studies will be presented with detailed description of snow surfaces characteristics, propulsions performance, the processes of interaction of the movers with the non-cohesive terrains, as well as advanced operating algorithms for unmanned vehicles.

For a realistic prediction of the agility and performance of vehicles, one needs to account for uncertainties. Such uncertainties can result from poorly known or variable parameters (e.g., variation in suspension stiffness and damping characteristics), from uncertain inputs (e.g., soil properties in vehicle-terrain interaction), or from rapidly changing forcings that can be best described in a stochastic framework (e.g., rough terrain profile).

This study presents a computationally efficient technique, which applies the generalized polynomial chaos theory to formally assess the uncertainty in multibody dynamic systems, with direct applications to vehicle systems. Moreover, the methodology is being applied to investigate the agility of vehicles operating in off-road conditions by developing stochastic tire models, stochastic tire-snow interaction models, stochastic terrain profile, and stochastic soil model equations. The approach used is to extend the model along the stochastic dimension to explicitly parameterize the uncertainty distribution. Polynomial chaos offers an efficient computational approach for the large, nonlinear multibody models of engineering systems of interest, such as off-road vehicles. For such systems, the number of uncertain parameters is relatively small, while the magnitude of uncertainties can be very large (e.g., vehicle-soil interaction). The methods discussed in this study allow the quantification of uncertainties and enable the simulations to produce results with “error bars”. Moreover, the proposed methodology allows the quantification of uncertainties in both time and frequency domains. It proposes direct statistical collocation as an alternative to the Galerkin method, and establishes its equivalence to a surface response approach.

One fundamental difficulty in understanding the physics of the off-road traction and in predicting vehicle performance is the indeterminacy of certain important parameters on the interface between tire and terrain/road surface, for instance, the slip ratio, the slip angle, the normal forces, and the friction coefficients. It is not possible to accurately capture the effect of such uncertainties on the tire behavior (resultant force and moments) using a deterministic model. The development of stochastic tire models with uncertain stiffness and damping characteristics will be illustrated on an original semi-analytical model based on a flexible band approach developed in Dr. Sandu’s research group. Moreover, two of the most popular semi-empirical models for predicting the pneumatic tires performance under steady-state and transient conditions, the Friction Ellipse Model and the Magic Formula Model, were selected to be extended from the deterministic to a stochastic workframe, to account for the uncertainties in the tire-terrain interface.

For the analysis to be complete, in off-road applications, one needs to investigate not only the vehicle and its interaction with the terrain, but also to model the terrain profile and the soil/snow/ice properties. Thus, the development (from a limited amount of test data) of a 3D terrain profile that preserves the stochastic properties of the real terrain measured will be presented. The technique constructs the virtual proving ground as stochastic partial differential equations. Moreover, the methodology to transform deterministic soil models into stochastic once will also be outlined. The soil properties are then integrated with the terrain profile for a complete off-road proving ground.

For all-wheel drive (AWD) terrain vehicles, along with noise and exhaust emissions, the level of the harmful vehicle impact on soil is another important environmental factor. This presentation introduces new methods to reduce the environmental damage caused by multi-wheel vehicle platforms.

The assessment of the destructive impact of the vehicle on soil is based on a method, which takes into account the decreased reproductive capability of the soil due to passage of a wheel, as compared with its natural capability. This method provides a comparative evaluation of various vehicles, both manufactured and designed ones and allows optimizing the design of AWD terrain vehicles.

The presentation is specifically concentrated on the intelligent automatic drivelines that are able to distribute engine power between the drive wheels. These drivelines have the advantages not only to improve vehicle performance (including passability, vehicle dynamics, fuel consumption) but also able to reduce the environmental damage. Relationships defining the optimum power distribution between the drive wheels of the AWD vehicle are discussed. Design principles and schemes of the intelligent driveline control system are presented.

A new software product for ecological safety assessment and optimization of the vehicle design is also presented. It allows to optimize the number of axles and their arrangement along the wheel-base; the distribution of load between the axles; the type and characteristics of the driveline; tire models and pressure; turning diagrams, etc. The developed computer product also allows controlling real-time dynamics of multi-wheel vehicles in terrain conditions.


SESSION 5: Modeling and Design of Vehicle Agile Systems
Session Chair: Anton Tumasov, Nizhny Novgorod State Technical University n.a. R.E. Alekseev

This presentation considers the computer modeling and simulation of vehicle dynamics with a particular focus on agility. Aspects of agile vehicle dynamics are considered initially in terms of the performance metrics that define it, the vehicle design parameters that influence it and the vehicle and tire modeling/simulation methodologies to investigate it.

Agility in a vehicle dynamics context is considered from the driving command task perspective and the requirement to deliver maximum rate of change of path curvature. Parameters such as mass, mass distribution, wheel base and tire size and stiffness are amongst those discussed here.

Modeling and simulation is covered to include a range of methodologies from simple two degree of freedom mathematical representations of the vehicle that can be used readily to investigate the sensitivity of lateral acceleration to basic vehicle parameters through to detailed multi-body models including mechanical representations of suspension, steering and powertrain systems that can be used for on and off road simulation.

Tire modeling is an important aspect of any simulation strategy in vehicle dynamics and the presentation will cover the essential theories associated with the most significant behaviors in the tire contact patch to provide and understanding of key parameters such as cornering stiffness, slip ratio and pneumatic trail. The discussion here covers implementations from a simple representation of cornering stiffness, empirical models such as the Magic Formula that fit parameters to measured tire test data obtained from tire test machines and increasingly from actual instrumented vehicles through to full physical finite element models of tires. The use of finite element modeling of tires is discussed in terms of the computationally intensive potential to model vehicles traversing terrain such as those including soft soils. The FTIRE model based on flexible elements is also covered in this presentation together with its use to predict load paths for durability assessment of vehicles operating in harsh environments.

The critical issues related to design, development and control of advanced suspensions, which have received increased attention in the recent years, are addressed. Various classes of controllable suspension systems are presented, including Magneto rheological (MR) dampers and electromagnetic suspensions. The design issues related to designing an advanced controllable suspension is evaluated in terms of the system viability, practicality, and durability. Various methods of controlling an active or semiactive suspension is addressed, along with the common methods that are used in both systems currently in existence and those in research and development stage for future implementation. The effect of advanced suspensions on vehicle handling and agility is discussed in detail for both on-road and off-road vehicles, as it relates to passenger cars, semi-trucks, and military vehicles. The discussions are intended to highlight not only the benefits of advanced suspensions for various classes of vehicles, but also the challenges they hold during the transformation from R&D to commercial implementation.

Traffic crashes caused 33,800 fatalities and 2.2 million injuries in the United States during 2009, resulting in an estimated economic loss of $230.6 billion. This talk covers some critical research on the pervasive vehicle safety problem. First, a holistic approach to vehicular safety and its advanced research challenges is discussed. Second, a detailed example of a specific complex driver assistance system for detection of fatigued and drowsy driving is presented. Third, related research in control systems for semi-autonomous driving and evasive maneuvers are reviewed.

Integrated vehicle passive and active safety systems are required to mitigate crashes or avoid collisions. In crashworthiness research, computational mechanics methods are used for the modeling, simulation, and design of passive safety systems. Methods in vehicle mechanics/dynamics, controls, communications, man-machine interface, human factors, as well as cognitive science are utilized to design and develop active safety systems such as warning methods, driver assistance, and automatic collision avoidance. First, some advances in both crashworthiness and collision avoidance methods are reviewed. Next, as an example of an advanced driver assistance system, development of a novel signal processing and pattern recognition method for unobtrusive detection of drowsy and fatigued drivers is presented. The algorithm uses an Empirical Mode Decomposition of the steering angle signal to extract specific drowsiness-affected features. A pattern recognition scheme is devised to classify the measured features into alert or drowsy states. The method is independent of road geometry and lane markings and accounts for driver variability. Our licensed and patent pending algorithm has shown promising results in driving simulator experiments, with over 80% accuracy in detecting drowsiness sufficiently early, prior to the emergence of impending hazards (i.e. crashes or lane departures). This talk concludes with a brief overview of control methods for semi-autonomous vehicle steering, which are essential for the next generation of driver assistance systems.