Terrain classification using ToF sensors for the enhancement of agricultural machinery traversability

Ground properties influence various aspects of mobile machinery navigation including localization, mobility status or task execution. Excessive slipping, skidding or trapping situations can compromise the vehicle itself or other elements in the workspace. Thus, detecting the soil surface characteristics is an important issue for performing different activities in an efficient, safe and satisfactory manner. In agricultural applications, this point is specially important since activities such as seeding, fertilizing, or ploughing are carried on within off-road landscapes which contain a diversity of terrains that modify the navigation behaviour of the vehicle. Thus, the machinery requires a cognitive capability to understand the surrounding terrain type or its characteristics in order to take the proper guidance or control actions. This work is focused on the soil surface classification by implementing a visual system capable to distinguish between five usual types of off-road terrains. Computer vision and machine learning techniques are applied to characterize the texture and color of images acquired with a Microsoft Kinect V2 sensor. In a first stage, development tests showed that only infra-red and RGB streams are useful to obtain satisfactory accuracy rates (above 90%). The second stage included field trials with the sensor mounted on a mobile robot driving through various agricultural landscapes. These scenarios did not present illumination restrictions nor ideal driving roads; hence, conditions can resemble real agricultural operations. In such circumstances, the proposed approach showed robustness and reliability, obtaining an average of 85.20% of successful classifications when tested along 17 trials within agricultural landscapes.     

On the impact of cargo weight, vehicle parameters, and terrain characteristics on the prediction of traction for off-road vehicles

A realistic prediction of the traction capacity of vehicles operating in off-road conditions must account for stochastic variations in the system itself, as well as in the operational environment. Moreover, for mobility studies of wheeled vehicles on deformable soil, the selection of the tire model used in the simulation influences the degree of confidence in the output. Since the same vehicle may carry various loads at different times, it is also of interest to analyze the impact of cargo weight on the vehicle’s traction.This study focuses on the development of an algorithm to calculate the tractive capacity of an off-road vehicle with stochastic vehicle parameters (such as suspension stiffness, suspension damping coefficient, tire stiffness, and tire inflation pressure), operating on soft soil with an uncertain level of moisture, and on a terrain topology that induces rapidly changing external excitations on the vehicle. The analysis of the vehicle–soil dynamics is performed for light cargo and heavy cargo scenarios. The algorithm relies on the comparison of the ground pressure and the calculated critical pressure to decide if the tire can be approximated as a rigid wheel or if it should be modeled as a flexible wheel. It also involves using previously-developed vehicle and stochastic terrain models, and computing the vehicle sinkage, resistance force, tractive force, drawbar pull, and tractive torque.The vehicle model used as a case study has seven degrees of freedom. Each of the four suspension systems is comprised of a nonlinear spring and a viscous (linear or magneto-rheological) damper. An off-road terrain profile is simulated as a 2-D random process using a polynomial chaos approach [Sandu C, Sandu A, Li L. Stochastic modeling of terrain profiles and soil parameters. SAE 2005 transactions. J Commer Vehicles 2005-01-3559]. The soil modeling is concerned with the efficient treatment of the impact of the moisture content on relationships critical in defining the mobility of an off-road vehicle (such as the pressure–sinkage [Sandu C et al., 2005-01-3559] and the shear stress–shear displacement relations). The uncertainties in vehicle parameters and in the terrain profile are propagated through the vehicle model, and the uncertainty in the output of the vehicle model is analyzed [Sandu A, Sandu C, Ahmadian M. Modeling multibody dynamic systems with uncertainties. Part I: theoretical and computational aspects, Multibody system dynamics. Publisher: Springer Netherlands; June 29, 2006. p. 1–23 (23), ISSN: 1384-5640 (Paper) 1573-272X (Online). doi:10.1007/s11044-006-9007-5; Sandu C, Sandu A, Ahmadian M. Modeling multibody dynamic systems with uncertainties. Part II: numerical applications. Multibody system dynamics, vol. 15, No. 3. Publisher: Springer Netherlands; 2006. p. 241–62 (22). ISSN: 1384-5640 (Paper) 1573-272X (Online). doi:10.1007/s11044-006-9008-4]. Such simulations can provide the basis for the study of ride performance, handling, and mobility of the vehicle in rough off-road conditions.     

Prediction of impacts of wheeled vehicles on terrain

Traffic of off-road vehicles can disturb soil, decrease vegetation development, and increase soil erosion. Terrain impacts caused by wheeled off-road vehicles were studied in this paper. Models were developed to predict terrain impacts caused by wheeled vehicles in terms of disturbed width and impact severity. Disturbed width and impact severity are not only controlled by vehicle types and vehicle dimensions, but also influenced by soil conditions and vehicle dynamic properties (turning radius, velocity). Field tests of an eight-wheeled vehicle and a four-wheeled vehicle were conducted to test these models. Field data of terrain–vehicle interactions in different vehicle dynamic conditions were collected. Vehicle dynamic properties were derived from a global position system (GPS) based tracking system. The average prediction percentage error of the theoretical disturbed width model is less than 20%. The average absolute error between the predicted impact severity and the measured value is less than an impact severity value of 12%. These models can be used to predict terrain impacts caused by off-road wheeled vehicles.     

The effect of velocity on rigid wheel performance

A simulation model to predict the effect of velocity on rigid-wheel performance for off-road terrain was examined. The soil–wheel simulation model is based on determining the forces acting on a wheel in steady state conditions. The stress distribution at the interface was analyzed from the instantaneous equilibrium between wheel and soil elements. The soil was presented by its reaction to penetration and shear. The simulation model describes the effect of wheel velocity on the soil–wheel interaction performances such as: wheel sinkage, wheel slip, net tractive ratio, gross traction ratio, tractive efficiency and motion resistance ratio. Simulation results from several soil-wheel configurations corroborate that the effect of velocity should be considered. It was found that wheel performance such as net tractive ratio and tractive efficiency, increases with increasing velocity. Both, relative wheel sinkage and relative free rolling wheel force ratio, decrease as velocity increases. The suggested model improves the performance prediction of off-road operating vehicles and can be used for applications such as controlling and improving off-road vehicle performance.     

Off-road tyre modelling II: effect of camber on tyre performance

The problem of off-road vehicle tyre-terrain interaction is that it is difficult to model accurately. For an off-road vehicle over medium to firm terrain, the tyre load may be entirely supported by the tips of the lugs, or with a minimum carcass contact with the terrain. In this case, the effect of the lugs should be taken into consideration. The forces at the interface between lugged tyre and the soil, including normal and shear stresses, are discussed in this paper. The multi-spoke tyre model was developed to study the effect of tyre lugs on the forces between tyre and terrain and it has been extended to predict the tyre forces and moments in the case of combined lateral and longitudinal slip for a cambered tyre. The influence of slip angle, camber angle and soil hardness on off-road tyre performance has been investigated. A computer program was developed using MATLAB software. The results were derived as tyre forces and moments in the three directions along the tyre contact length. A comparison between the results of the multi-spoke tyre model of a smooth off-road tyre and an off-road tyre with straight lugs, in the cambered case, has been made. The results indicated that slip angle, camber angle and soil characteristics have a strong effect on off-road tyre performance. The modified mathematical model results help the off-road tyre engineering designers to predict accurate values of tyre forces and moments in this complex case.     

Parameterization and modelling of large off-road tyres for ride analyses: Part 2 – Parameterization and validation of tyre models

Every mathematical model used in a simulation is an idealization and simplification of reality. Vehicle dynamic simulations that go beyond the fundamental investigations require complex multi-body simulation models. The tyre–road interaction presents one of the biggest challenges in creating an accurate vehicle model. Many tyre models have been proposed and developed but proper validation studies are less accessible. These models were mostly developed and validated for passenger car tyres for application on relatively smooth roads. The improvement of ride comfort, safety and structural integrity of large off-road vehicles, over rough terrain, has become more significant in the development process of heavy vehicles. This paper investigates whether existing tyre models can be used to accurately describe the vertical behaviour of large off road tyres while driving over uneven terrain. [1] Presented an extensive set of experimentally determined parameterization and validation data for a large off-road tyre. Both laboratory and field test are performed for various loads, inflation pressures and terrain inputs. The parameterization process of four tyre models or contact models are discussed in detail. The parameterized models are then validated against test results on various hard but rough off-road terrain and the results are discussed.     

A roll stability performance measure for off-road vehicles

The roll stability is significant for both road and off-road commercial vehicles, while the majority of reported studies focus on road vehicles neglecting the contributions of uneven off-road terrains. The limited studies on roll stability of off-road vehicles have assessed the stability limits using performance measures derived for road vehicles. This study proposes an alternative performance measure for assessing roll stability limits of off-road vehicles. The roll dynamics of an off-road mining vehicle operating on random rough terrains are investigated, where the two terrain-track profiles are synthesized considering coherency between them. It is shown that a measure based on steady-turning root-mean-square lateral acceleration corresponding to the sustained period of unity lateral-load-transfer-ratio prior to the absolute-rollover, could serve as a reliable measure of roll stability of the vehicle operating on random rough terrains. The robustness of proposed performance measure is demonstrated considering sprung mass center height variations and different terrain excitations. The simulation results revealed adverse effects of terrain elevation magnitude on the roll stability, while a relatively higher coherency resulted in lower terrain roll-excitation and thereby enhanced vehicle roll stability. Terrains with relatively higher waviness increased the magnitude of lower spatial frequency components, which resulted in reduced roll stability limits.     

Finite element modeling of tire/terrain interaction: Application to predicting soil compaction and tire mobility

Tire/terrain interaction has been an important research topic in terramechanics. For off-road vehicle design, good tire mobility and little compaction on terrain are always strongly desired. These two issues were always investigated based on empirical approaches or testing methods. Finite element modeling of tire/terrain interaction seems a good approach, but the capability of the finite element has not well demonstrated. In this paper, the fundamental formulations on modeling soil compaction and tire mobility issues are further introduced. The Drucker-Prager/Cap model implemented in ABAQUS is used to model the soil compaction. A user subroutine for finite strain hyperelasticity model is developed to model nearly incompressible rubber material for tire. In order to predict transient spatial density, large deformation finite element formulation is used to capture the configuration change, which combines with soil elastoplastic model to calculate the transient spatial density due to tire compaction on terrain. Representative simulations are provided to demonstrate how the tire/terrain interaction model can be used to predict soil compaction and tire mobility in the field of terramechanics.     

Modelling of off-road wheeled vehicles for real-time dynamic simulation

Simulation of wheel-ground and vehicle-ground interactions is very important in many applications. Achieving accuracy and efficiency is challenging for both soft and hard terrains. This is not only because of the simulation and numerical challenges, but also due to the questionable nature of the existing terrain models. For example, the most widely used terramechanics model is not a representative constitutive relation for a full range of dynamic conditions and applications, but rather a parametrization of steady state conditions. In general, the selection and development of the proper constitutive model and the parametrization of the ground properties are very challenging. Here, we present a unified framework for general wheel-ground interaction which can be used with different terramechanics models. The framework is based on a complementarity formulation and also uses the concept of kinematic constitutive relations, beside the other known concepts for modelling and parametrizing the soil properties. The framework makes it possible to consider the appropriate modelling of the terrain for a broad range of dynamic behaviours and simulation conditions. We will illustrate the material with several examples for off-road conditions.     

Motion resistance measurements on large lug tyres

Motion resistance of tyres directly contribute to the operational costs of all vehicles. Advances in the design and simulation of large off-road vehicles (construction, mining, agriculture etc.) have increased the need for accurate models of large off-road tyres. Vehicle OEMs use coast down and drawbar pull tests to determine the motion resistance of tyres used. Drum test rigs and motion resistance test trailers can also be used to determine motion resistance. Most research on motion resistance to date have been conducted on passenger car tyres with on-road truck tyres coming into focus. Motion resistance studies on agricultural tyres traversing over deformable terrain have been conducted in the past. However as more off-road vehicle are being used on-road OEMs of off-road vehicle are infesting in motion resistance measurements on non-deformable terrain. This paper compares different methods used to measure the motion resistance of a large lug tyre, as used in agricultural applications, on non-deformable terrain. Some basic considerations that need to be taken into account are the very low longitudinal forces that need to be measured compared to the large vertical load carried by the tyre and tyre operating conditions.     

Calculating fractal parameters from low-resolution terrain profiles

Driver comfort on rough terrain is an important factor in the off-road performance of wheeled and tracked ground vehicles. The roughness of a terrain has typically been quantified by the U.S. Army as the root-mean-square elevation deviation (RMS) of the terrain profile. Although RMS is an important input into many mobility calculations, it is not scale invariant, making it difficult to estimate RMS from low resolution terrain profiles. Fractal parameters are another measure of roughness that are scale invariant, making them a convenient proxy for RMS. While previous work found an empirical relationship between fractal dimension and RMS, this work will show that, by including the cutoff length, an analytic relationship between fractal properties and RMS can be employed. The relationship has no free parameters and agrees very well with experimental data - thus providing a powerful predictive tool for future analyses and a reliable way to calculate surface roughness from low-resolution terrain data in a way that is scale invariant. In addition, we show that this method applies to both man-made ride courses and natural terrain profiles.     

Analysis of vehicle platoon movement and speed-spacing relationships during military exercises

In this study a method that identifies off-road vehicle column movement was developed and evaluated. Previous studies have revealed that multiple vehicle passes produce detrimental soil and terrain impacts. Identifying the frequency and location of this type of multi-pass impact during military maneuvers is difficult. This method will aid in the assessment of environmental impacts of off-road military vehicles by allowing land managers to characterize vehicle movement patterns, especially column movement, at military training installations during maneuvers. GPS units mounted on military vehicles collected on and off-road tracking data during a reconnaissance maneuver at Fort Lewis Military Installation, Washington. A set of data utilizing a Stryker platoon of four vehicles was used to evaluate this method. The GPS coordinates, speed, and direction of travel of each vehicle was collected at each second. A criteria to identify platoon column movement was developed based on vehicle proximity, speed and direction of travel. The results of this study show that the method can correctly identify off-road column movement for the purpose of evaluating the multi-pass impacts on the terrain. In addition, using this approach the vehicle movement patterns associated with on- and off-road platoon movement (i.e. vehicle speeds and spacing) were evaluated.     

Dynamics modeling and robotic-assist,leader-follower control of tractor convoys

This paper proposes a generalized dynamics model and a leader-follower control architecture for skid-steered tracked vehicles towing polar sleds. The model couples existing formulations in the literature for the powertrain components with the vehicle-terrain interaction to capture the salient features of terrain trafficability and predict the vehicles response. This coupling is essential for making realistic predictions of the vehicles traversing capabilities due to the power-load relationship at the engine output. The objective of the model is to capture adequate fidelity of the powertrain and off-road vehicle dynamics while minimizing the computational cost for model based design of leader-follower control algorithms. The leader-follower control architecture presented proposes maintaining a flexible formation by using a look-ahead technique along with a way point following strategy. Results simulate one leader-follower tractor pair where the leader is forced to take an abrupt turn and experiences large oscillations of its drawbar arm indicating potential payload instability. However, the follower tractor maintains the flexible formation but keeps its payload stable. This highlights the robustness of the proposed approach where the follower vehicle can reject errors in human leader driving.     

Torque distribution influence on tractive efficiency and mobility of off-road wheeled vehicles

Off-road vehicle performance is strongly influenced by the tire-terrain interaction mechanism. Soft soil reduces traction and significantly modifies vehicle handling; therefore tire dynamics plays a strong role in off-road mobility evaluation and needs to be addressed with ad-hoc models. Starting from a semi-empirical tire model based on Bekker–Wong theory, this paper, analyzes the performance of a large four wheeled vehicle driving on deformable terrain. A 14 degree of freedom vehicle model is implemented in order to investigate the influence of torque distribution on tractive efficiency through the simulation of front, rear, and all wheel drive configuration. Results show that optimal performance, regardless vertical load distribution, is achieved when torque is biased toward the rear axle. This suggests that it is possible to improve tractive efficiency without sacrificing traction and mobility. Vehicle motion is simulated over dry sand, moist loam, flat terrain and inclined terrain.     

A comparison of ground vehicle mobility analysis based on soil moisture time series datasets from WindSat,LIS, and in situ sensors

Soil moisture is a key terrain variable in ground vehicle off-road mobility. Historically, models of the land water balance have been used to estimate soil moisture. Recently, satellites have provided another source of soil moisture estimates that can be used to estimate soil-limited vehicle mobility. In this study, we compared the off-road vehicle mobility estimates based on three soil moisture sources: WindSat (a satellite source), LIS (a computer model source), and in situ ground sensors (to represent ground truth). Mobility of six vehicles, each with different ranges of sensitivity to soil moisture, was examined in three test sites. The results demonstrated that the effect of the soil moisture error on mobility predictions is complex and may produce very significant errors in off-road mobility analysis for certain combinations of vehicles, seasons, and climates. This is because soil moisture biases vary in both direction and magnitude with season and location. Furthermore, vehicles are sensitive to different ranges of soil moistures. Modeled vehicle speeds in the dry time periods were limited by the interaction between soil traction and the vehicles’ powertrain characteristics. In the wet season, differences in soil strength resulted in more significant differences in mobility predictions.     

Comparative study on multibody vehicle dynamics models based on subsystem synthesis method using Cartesian and joint coordinates

The subsystem synthesis method has been developed in order to improve computational efficiency for a multibody vehicle dynamics model. Using the subsystem synthesis method, equations of motion of the base body and each subsystem can be solved separately. In the subsystem synthesis method, various coordinate systems can be used and various integration methods can be applied in each subsystem, as long as the effective mass matrix and the effective force vector are properly produced. In this paper, comparative study has been carried out for the subsystem synthesis method with Cartesian coordinates and with joint relative coordinates. Two different integration methods such as an explicit integrator and an explicit implicit integrator are employed. In order to see the accuracy and computational efficiency from the different models based on the different coordinate systems and different integration methods, a rough terrain run simulations has been carried out with a 6 × 6 off-road multibody vehicle model.     

Recurrent and convolutional neural networks for deep terrain classification by autonomous robots

The future challenge for field robots is to increase the level of autonomy towards long distance (>1 km) and duration (>1h) applications. One of the key technologies is the ability to accurately estimate the properties of the traversed terrain to optimize onboard control strategies and energy efficient path-planning, ensuring safety and avoiding possible immobilization conditions that would lead to mission failure. Two main hypotheses are put forward in this research. The first hypothesis is that terrain can be effectively detected by relying exclusively on the measurement of quantities that pertain to the robot-ground interaction, i.e., on proprioceptive signals. Therefore, no visual or depth information is required. Then, artificial deep neural networks can provide an accurate and robust solution to the classification problem of different terrain types. Under these hypotheses, sensory signals are classified as time series directly by a Recurrent Neural Network or by a Convolutional Neural Network in the form of higher-level features or spectrograms resulting from additional processing. In both cases, results obtained from real experiments show comparable or better performance when contrasted with standard Support Vector Machine with the additional advantage of not requiring an a priori definition of the feature space.     

Application of the computer simulation model NTVPM-86 to the development of a new version of the infantry fighting vehicle ASCOD

In the past, the task of evaluating soft-ground mobility of off-road vehicles has been carried out primarily using empirical methods (or models), such as the NATO Reference Mobility Model (NRMM) or the Rowland method based on the mean maximum pressure (MMP). The databases for these empirical methods were mostly established decades ago. Consequently, in many cases, they cannot be used in evaluating new generations of vehicles with new design features, as the mobility of these vehicles simply cannot be described within the limits of these empirical databases.Since the 1980s, a series of comprehensive and realistic simulation models for design and performance evaluation of off-road vehicles has emerged. They are based on the detailed studies of the physical nature of vehicle-terrain interaction, taking into account all major vehicle design features and pertinent terrain characteristics. This paper describes the application of one of these models, known as NTVPM-86, developed by Vehicle Systems Development Corporation, Canada, to the design and development of a new version of the ASCOD infantry fighting vehicle, produced by a joint venture formed by Empresa Nacional Santa Barbara of Spain and Steyr-Daimler-Puch of Austria. The results of field tests performed by the Military Technology Agency, Ministry of Defence, Vienna, Austria and released recently confirm that, as predicted by the NTVPM-86 model, the new version of the ASCOD has much improved performance than the original over soft terrain, including soft clay and snow-covered terrain. This is another example of the successful application of the NTVPM-86 model to the design and development of a new generation of high-speed tracked vehicles.     

Maneuver analysis methodology to predict vehicle impacts on training lands

Tactical mobility analysis techniques were merged with land management strategies to assess potential impacts of vehicle operations on training areas for rangeland planning and management. A vehicle mobility analysis was performed for a suite of vehicle types using the NATO Reference Mobility Model (NRMM II). Input parameters include terrain information (soil type, slope, vegetation, surface roughness, soil strength), terrain surface condition based on climate (terrain strength, freeze–thaw, moisture content, snow cover), and vehicle specifications (tire, power train, weight on each axle, ground clearance, dimensions, ride). The vehicle performance was spatially mapped over the terrain for different seasons of the year and used to calculate the maneuverable acreage, which was compared to acreage needed for training requirements. This can be related to land capability based on expected training impact (Maneuver Impact Miles, MIM) and Land Condition Curves which link training density to land condition. This methodology can be used to determine the suitability of training lands and the degree of land management or rehabilitation expected. The methodology was applied to the transformation of the Alaska training lands to support a new brigade unit called the Stryker Brigade Combat Team (SBCT3), but is equally useful for other training areas and military units. For summer use, Alaska training lands are capable of supporting four times the projected training requirements. For winter, when the ground is frozen, more than 10 times the area needed was available.