This innovative practice full paper describes a new software framework based on JU nit to test student work. Automated testing is an important capability when teaching software development at the college level. Ideally, a testing system will allow the instructor to efficiently create a thorough set of tests. Also, the software should facilitate grading tasks and produce informative reports that can be distributed to the students in a timely fashion. For Java development, the well-known JUnit framework enables a test suite to be applied to a student's submission. The mutools library presented here extends the JUnit framework in novel ways to accelerate the instructor's task of creating test suites. This new framework allows the instructor to augment tests with directives to control scoring and reporting. The four main capabilities of the software include: 1) An assert statement that does not terminate the test when it has failed. Instead, statistics are maintained regarding the success or failure of each assert statement. 2) Tests that can be configured to award partial credit. This can be useful in situations where the instructor deems it appropriate to award students some credit even in the presence of incorrect asserts. 3) Tests that can be grouped into categories that match a particular rubric item. Java annotations are placed on the test suite to define these categories. For example, @TestCategory(name= “remove”, points=10.0) specifies that 10 points will be awarded for successful implementation of all tests related to removing an item from the collection class. 4) Testing reports that contain varying levels of information. With minimal details, the testing report shows each testing category with the following information: assert statistics, whether the test timed out or had abnormal termination. This software has been used for many semesters and has been found to increase the speed at which the instructor can develop test suites for grading. The framework is available on github.

Head-mounted display (HMD) based immersive virtual environment (VE) systems that incorporate a wearable rendering unit allow users to navigate within virtual worlds through natural walking. Redirected walking (RDW) is a technique that allows users to explore virtual worlds which are larger than the physical tracking area. It involves imperceptibly rotating the VE that the user sees, which causes the user to subconsciously compensate by physically turning. This work extends generalized RDW techniques to allow two immersed users to share a tracking area. The extension forecasts potential collisions so they can be avoided by using RDW techniques. In simulations based on recorded user-data, unsafe situations which could result in a collision occurred at a rate of 31.5/hr. in baseline experiments. The algorithm presented here resulted in all potential collisions being predicted in advance. Once predicted, most future collisions could be avoided using RDW techniques. Some, however, were resolved by stopping one or both users. These instances occurred at a rate of only 2.5/hr.

A goal of redirected walking (RDW) is to allow large virtual worlds to be explored within small tracking areas. Generalized steering algorithms, such as steer-to-center, simply move the user toward locations that are considered to be collision free in most cases. The algorithm developed here, FORCE, identifies collision-free paths by using a map of the tracking area's shape and obstacles, in addition to a multistep, probabilistic prediction of the user's virtual path through a known virtual environment. In the present implementation, the path predictions describe a user's possible movements through a virtual store with aisles. Based on both the user's physical and virtual location / orientation, a search-based optimization technique identifies the optimal steering instruction given the possible user paths. Path prediction uses the map of the virtual world; consequently, the search may propose steering instructions that put the user close to walls if the user's future actions eventually lead away from the wall. Results from both simulated and real users are presented. FORCE identifies collision-free paths in 55.0 percent of the starting conditions compared to 46.1 percent for generalized methods. When considering only the conditions that result in different outcomes, redirection based on FORCE produces collision-free path 94.5 percent of the time.

A hybrid evolutionary learning algorithm is presented that synthesizes a complete multiclass pattern recognition system. The approach uses a multifaceted representation that evolves layers of processing to perform feature extraction from raw input data, select cooperative sets of feature detectors, and assemble a linear classifier that uses the detectors' responses to label targets. The hybrid algorithm, called hybrid evolutionary learning for pattern recognition (HELPR), blends elements of evolutionary programming, genetic programming, and genetic algorithms to perform a search for an effective set of feature detectors. Individual detectors are represented as expressions composed of morphological and arithmetic operations. Starting with a few small random expressions, HELPR expands the number and complexity of the features to produce a recognition system that achieves high accuracy. Results are presented that demonstrate the performance of HELPR-generated recognition systems applied to the task of classification of high-range resolution radar signals.

This correspondence presents two soft morphological algorithms that process multiple images simultaneously. The first algorithm performs best when the structuring elements contain less than 19 points; whereas, the second algorithm should be used for larger structuring elements. Theoretical and experimental analyses show these algorithms are faster than the conventional algorithm.

This paper describes a system based on evolutionary learning, called MORPH, that semi-automates the generation of morphological programs. MORPH maintains a population of morphological programs that is continually enhanced. The first phase of each learning cycle synthesizes morphological sequences that extract novel features which increase the population's diversity. The second phase combines these newly formed operator sequences into larger programs that are better than the individual programs. A stochastic selection process eliminates the poor performers, while the survivors serve as the basis of another learning cycle. Experimental results are presented for binary and grayscale target recognition problems.