Search Results

Using neural networks, an algorithm has been developed to steer a wheel loader vehicle. Mathematical functions have been used in the past in an attempt to model a human in their operation of many types of vehicles. Since such functions can typically only be derived for situations in which the problem domain is thoroughly understood, research continues in an effort to develop a complete “operator model”. Neural Network algorithms were utilized in an attempt to determine the feasibility of accurately modeling the operator of a wheel loader construction vehicle. These algorithms were also used to determine how the control of different vehicle functions might be automated on a wheel loader.

This research compares the responses of a vehicle modeled in the 3D vehicle simulation program SIMON in the HVE simulation operating system against instrumented responses of a 3-axle tractor, 2-axle semi-trailer combination. The instrumented tests were previously described in SAE 2001-01-0139 and SAE 2003-01-1324 as part of a continuous research effort in the area of vehicle dynamics undertaken at the Vehicle Research and Test Center (VRTC). The vehicle inertial and mechanical parameters were measured at the University of Michigan Transportation Research Institute (UMTRI). The tire data was provided by Smithers Scientific Services, Inc. and UMTRI. The series of tests discussed herein compares the modeled and instrumented vehicle responses during quasi-steady state, steady state and transient handling maneuvers, producing lateral accelerations ranging nominally from 0.05 to 0.5 G's.

In this paper we demonstrate our work to date on our underactuated two link robot called the Pendubot. First we will overview the Pendubot's design, discussing the components of the linkage and the interface to the PC making up the controller. Parameter identification of the Pendubot is accomplished both by solid modeling methods and energy equation least squares techniques. With the identified parameters, mathematical models are developed to facilitate controller design. The goal of the control is to swing the Pendubot up and balance it about various equilibrium configurations. Two control algorithms are used for this task. Partial feedback linearization techniques are used to design the swing up control. The balancing control is then designed by linearizing the dynamic equations about the desired equilibrium point and using LQR or pole placement techniques to design a stabilizing controller.

Virtual reality technology permits engineers to assess visibility of virtual machines interactively in a virtual environment. Use of a virtual prototyping system allows one to investigate, in three dimensions, the effect of design changes on the visibility of critical machine components, such as the working tool. Although there have been successful applications of virtual prototyping systems to visibility assessment, there still exist system limitations that need to be overcome.

Intelligent convoy consisting of heavy duty vehicles is an implementation of IVHS believed to be one of the most practicable proposals to come into reality in the near feature. Control Strategy in the context of Autonomous Intelligent Vehicle Platoon is different from that in other “Lane-keeping” IVHS systems which have been well studied. In this paper, an Autonomous Platoon consisting two articulated commercial vehicles is studied and a model of tractor-trailer type commercial vehicles suitable for control studies is derived based on a single track three-axle bicycle model. The authors give perspectives on the implementation of intelligent convoy of articulated vehicles emphasizing safety issues in emergency situations, as opposed to normal following of the lead vehicle. An initial integrated braking and steering control is developed to avoid spinout or jack-knifing when specific axles are locked during braking process.

An overview of enrollment trends and curricular changes in engineering education in the past ten years. Comments are made about the implications of lower enrollment on quality of education and availability of engineers for the employment market. Discussions of curricular variations summarizes changes such as computerization of engineering studies, expansion of high school preparation, and selection of major studies for students.

Yield potential was predicted and mapped for three corn fields in Central Illinois, using digital aerial color infrared images. Three methods, namely statistical (regression) modeling, genetic algorithm optimization and artificial neural networks, were used for developing yield models. Two image resolutions of 3 and 6 m/pixel were used for modeling. All the models were trained using July 31 image and tested using images from July 2 and August 31, all from 1998. Among the three models, artificial neural networks gave best performance, with a prediction error less than 30%. The statistical model resulted in prediction errors in the range of 23 to 54%. The lower resolution images resulted in better prediction accuracy compared to resolutions higher than or equal to the yield resolution. Images after pollination resulted in better accuracy compared to images before pollination.