Terrain Characterization via Machine vs. Deep Learning Using Remote Sensing.

Terrain traversability is critical for developing Go/No-Go maps for ground vehicles, which significantly impact a mission's success. To predict the mobility of terrain, one must understand the soil characteristics. In-situ measurements performed in the field are the current method of collecting this information, which is time-consuming, costly, and can be lethal for military operations. This paper investigates an alternative approach using thermal, multispectral, and hyperspectral remote sensing from an unmanned aerial vehicle (UAV) platform. Remotely sensed data combined with machine learning (linear, ridge, lasso, partial least squares (PLS), support vector machines (SVM), and k nearest neighbors (KNN)) and deep learning (multi-layer perceptron (MLP) and convolutional neural network (CNN)) are used to perform a comparative study to estimate the soil properties, such as the soil moisture and terrain strength, used to generate prediction maps of these terrain characteristics. This study found that deep learning outperformed machine learning. Specifically, a multi-layer perceptron performed the best for predicting the percent moisture content (R2/RMSE = 0.97/1.55) and the soil strength (in PSI), as measured by a cone penetrometer for the averaged 0-6" (CP06) (R2/RMSE = 0.95/67) and 0-12" depth (CP12) (R2/RMSE = 0.92/94). A Polaris MRZR vehicle was used to test the application of these prediction maps for mobility purposes, and correlations were observed between the CP06 and the rear wheel slip and the CP12 and the vehicle speed. Thus, this study demonstrates the potential of a more rapid, cost-efficient, and safer approach to predict terrain properties for mobility mapping using remote sensing data with machine and deep learning algorithms.

Modeling Human Steering Behavior in Haptic Shared Control of Autonomy-Enabled Unmanned Ground Vehicles.

OBJECTIVE: A human steering model for teleoperated driving is extended to capture the human steering behavior in haptic shared control of autonomy-enabled Unmanned Ground Vehicles (UGVs). BACKGROUND: Prior studies presented human steering models for teleoperation of a passenger-sized Unmanned Ground Vehicle, where a human is fully in charge of driving. However, these models are not applicable when a human needs to interact with autonomy in haptic shared control of autonomy-enabled UGVs. How a human operator reacts to the presence of autonomy needs to be studied and mathematically encapsulated in a module to capture the collaboration between human and autonomy. METHOD: Human subject tests are conducted to collect data in haptic shared control for model development and validation. The ACT-R architecture and two-point steering model used in the previous literature are adopted to predict the operator's desired steering angle. A torque conversion module is developed to convert the steering command from the ACT-R model to human torque input, thus enabling haptic shared control with autonomy. A parameterization strategy is described to find the set of model parameters that optimize the haptic shared control performance in terms of minimum average lane keeping error (ALKE). RESULTS: The model predicts the minimum ALKE human subjects achieve in shared control. CONCLUSIONS: The extended model can successfully predict the best haptic shared control performance as measured by ALKE. APPLICATION: This model can be used in place of human operators, enabling fully simulation-based engineering, in the development and evaluation of haptic shared control technologies for autonomy-enabled UGVs, including control negotiation strategies and autonomy capabilities.

Modeling Human Steering Behavior in Teleoperation of Unmanned Ground Vehicles With Varying Speed.

OBJECTIVE: This paper extends a prior human operator model to capture human steering performance in the teleoperation of unmanned ground vehicles (UGVs) in path-following scenarios with varying speed. BACKGROUND: A prior study presented a human operator model to predict human steering performance in the teleoperation of a passenger-sized UGV at constant speeds. To enable applications to varying speed scenarios, the model needs to be extended to incorporate speed control and be able to predict human performance under the effect of accelerations/decelerations and various time delays induced by the teleoperation setting. A strategy is also needed to parameterize the model without human subject data for a truly predictive capability. METHOD: This paper adopts the ACT-R cognitive architecture and two-point steering model used in the previous work, and extends the model by incorporating a far-point speed control model to allow for varying speed. A parameterization strategy is proposed to find a robust set of parameters for each time delay to maximize steering performance. Human subject experiments are conducted to validate the model. RESULTS: Results show that the parameterized model can predict both the trend of average lane keeping error and its lowest value for human subjects under different time delays. CONCLUSIONS: The proposed model successfully extends the prior computational model to predict human steering behavior in a teleoperated UGV with varying speed. APPLICATION: This computational model can be used to substitute for human operators in the process of development and testing of teleoperated UGV technologies and allows fully simulation-based development and studies.

A workload adaptive haptic shared control scheme for semi-autonomous driving.

Haptic shared control is used to manage the control authority allocation between a human and an autonomous agent in semi-autonomous driving. Existing haptic shared control schemes, however, do not take full consideration of the human agent. To fill this research gap, this study presents a haptic shared control scheme that adapts to a human operator's workload, eyes on road and input torque in real time. We conducted human-in-the-loop experiments with 24 participants. In the experiment, a human operator and an autonomy module for navigation shared the control of a simulated notional High Mobility Multipurpose Wheeled Vehicle (HMMWV) at a fixed speed. At the same time, the human operator performed a target detection task. The autonomy could be either adaptive or non-adaptive to the above-mentioned human factors. Results indicate that the adaptive haptic control scheme resulted in significantly lower workload, higher trust in autonomy, better driving task performance and smaller control effort.

Modeling Human Steering Behavior During Path Following in Teleoperation of Unmanned Ground Vehicles.

OBJECTIVE: This paper presents a behavioral model representing the human steering performance in teleoperated unmanned ground vehicles (UGVs). BACKGROUND: Human steering performance in teleoperation is considerably different from the performance in regular onboard driving situations due to significant communication delays in teleoperation systems and limited information human teleoperators receive from the vehicle sensory system. Mathematical models capturing the teleoperation performance are a key to making the development and evaluation of teleoperated UGV technologies fully simulation based and thus more rapid and cost-effective. However, driver models developed for the typical onboard driving case do not readily address this need. METHOD: To fill the gap, this paper adopts a cognitive model that was originally developed for a typical highway driving scenario and develops a tuning strategy that adjusts the model parameters in the absence of human data to reflect the effect of various latencies and UGV speeds on driver performance in a teleoperated path-following task. RESULTS: Based on data collected from a human subject test study, it is shown that the tuned model can predict both the trend of changes in driver performance for different driving conditions and the best steering performance of human subjects in all driving conditions considered. CONCLUSIONS: The proposed model with the tuning strategy has a satisfactory performance in predicting human steering behavior in the task of teleoperated path following of UGVs. APPLICATION: The established model is a suited candidate to be used in place of human drivers for simulation-based studies of UGV mobility in teleoperation systems.

Compliant contact versus rigid contact: A comparison in the context of granular dynamics.

We summarize and numerically compare two approaches for modeling and simulating the dynamics of dry granular matter. The first one, the discrete-element method via penalty (DEM-P), is commonly used in the soft matter physics and geomechanics communities; it can be traced back to the work of Cundall and Strack [P. Cundall, Proc. Symp. ISRM, Nancy, France 1, 129 (1971); P. Cundall and O. Strack, Geotechnique 29, 47 (1979)GTNQA80016-850510.1680/geot.1979.29.1.47]. The second approach, the discrete-element method via complementarity (DEM-C), considers the grains perfectly rigid and enforces nonpenetration via complementarity conditions; it is commonly used in robotics and computer graphics applications and had two strong promoters in Moreau and Jean [J. J. Moreau, in Nonsmooth Mechanics and Applications, edited by J. J. Moreau and P. D. Panagiotopoulos (Springer, Berlin, 1988), pp. 1-82; J. J. Moreau and M. Jean, Proceedings of the Third Biennial Joint Conference on Engineering Systems and Analysis, Montpellier, France, 1996, pp. 201-208]. The DEM-P and DEM-C are manifestly unlike each other: They use different (i) approaches to model the frictional contact problem, (ii) sets of model parameters to capture the physics of interest, and (iii) classes of numerical methods to solve the differential equations that govern the dynamics of the granular material. Herein, we report numerical results for five experiments: shock wave propagation, cone penetration, direct shear, triaxial loading, and hopper flow, which we use to compare the DEM-P and DEM-C solutions. This exercise helps us reach two conclusions. First, both the DEM-P and DEM-C are predictive, i.e., they predict well the macroscale emergent behavior by capturing the dynamics at the microscale. Second, there are classes of problems for which one of the methods has an advantage. Unlike the DEM-P, the DEM-C cannot capture shock-wave propagation through granular media. However, the DEM-C is proficient at handling arbitrary grain geometries and solves, at large integration step sizes, smaller problems, i.e., containing thousands of elements, very effectively. The DEM-P vs DEM-C comparison is carried out using a public-domain, open-source software package; the models used are available online.