Aidan F. Browne

A Versatile Approach for Teaching Autonomous Robot Control to Multi-Disciplinary Undergraduate and Graduate Students

Learning autonomous vehicle programming can be very challenging for students, especially when combined with robotic vehicle design and construction. This paper presents a methodology successfully used over the past six years to teach autonomy using a versatile platform built upon a commercially available product. A number of courses have been taught using the methodology at both the undergraduate and graduate levels. Students’ ability to successfully learn and produce a solution to an autonomous robot control challenge has shown to be very effective.

Modeling a novel laser localization system

A novel laser based localization system is evaluated through mathematical modeling and simulation. The system concept was originally conceived as part of a university design project for use in a mobile robot mining competition. The overall operation of the system is accomplished using just two external beacons to perform triangulation. These rotating beacons emit laser pulses containing angular information; a rotating detector on the vehicle captures and decodes this information. This received data is analyzed in real time to resolve position and orientation. Potential errors may result from inconsistencies in the conversion of the beacons angular direction to laser frequencies. This can be compounded with errors from resolving angular data from the laser frequencies received by the detector. The rotational frequency of both the beacons, the detector, and the laser beam width are all factors that affect the time of reception, which is another potential source of error and can be used for optimization. The purpose of the model is to analyze parameters such as these in order to evaluate the effectiveness of the localization system and exploit areas of optimization. The model simulates the localization of the robot in static and dynamic conditions to produce data accounting for error, hit time, and position update time. The results obtained are expected to provide insight towards future design improvements.

Development of a remote laboratory architecture for Mission Critical Operations instruction

In order to support the remote training of Mission Critical Operators, a standard interface and interconnection architecture had to be developed. While many others have created schemes for remote operation of equipment, many tend to either be customized to a target hardware, lack the requisite security protocols, or not be scalable. We have created such an architecture, leveraging existing commercial or open source hardware and software, and creating customizations where necessary to integrate the parts into an entire system. The result is an experience that connects students with industrially relevant hardware and software in a manner that approximates an industrial operator involved in remotely troubleshooting or configuring a networked remote resource.

Implementation of a Cartesian robot for remote Mission Critical Operator training

As part of ongoing work to create training for Mission Critical Operators, a new training system was developed offering multiple degrees of freedom (DOF), which was a desirable addition as part of a progression of training systems of increasing complexity. With the addition of multiple DOF, the user may be exposed to a more complex control environment presenting challenging tasks related to full range of motion on multiple axes. The electromechanical training system was developed following a previously developed standard allowing control by two separate controllers; a Micro800 series programmable logic controller (PLC) or a National Instruments myRIO. This gives the user an opportunity to learn two programming languages: ladder logic and LabVIEW. The training system includes 4 DOF and has free range of motion in the X, Y, and Z axis. This allows for multiple applications to be simulated. The overall approach is presented; the final designs are being made available under an open-source license.

Enhanced model for laser localization system

A novel method for localizing a mobile platform within a region of interest was previously conceived and modeled. The system relies on data received at the platform from two independent rotating laser beacons that encode their current rotational angle within the light beam. Using that data and trigonometric calculations, the location and orientation of the mobile platform can be derived in real time. Information gleaned from the previous model led to the desire to enhance the model before realizing the system in hardware. Certain parameter variations are of interest, but the original models simulation time is prohibitive to performing useful dynamic parametric analysis, as it take over twelve hours to produce a set of data. An enhanced approach using Hough transforms and predictive calculations has been created and is presented here. The enhanced model calculates the same data sets in under 15 seconds, greatly improving the ability to use it for parametric analysis. Preliminary results from the model are presented.

Vector-based robot obstacle avoidance using LIDAR and mecanum drive

With their reduction in cost, LIDAR sensors are starting to become ubiquitous on robotic platforms that need to self-navigate. Omni-directional robotic drive systems have matured in performance, including mecanum drive systems. Both technologies are vector based, which makes their pairing a simplifying solution for robot navigation, and specifically obstacle avoidance. This paper steps through the implementation of both systems onto a test platform and how they can be used in a vector-based avoidance algorithm. A robot is programed to follow along a wall (with doors, small columns and other features indenting and protruding from the wall). When the robot encounters an obstacle in its path above a pre-determined size, it takes action to avoid colliding with it. In a multi-step approach, the robot strafes at an angle to clear the object, travels past the object, and then strafes back towards the wall to resume its original path.

Area acoustic and electromagnetic emissions monitoring of 3-phase motors

Equipment malfunction in electric power plant may lead to catastrophic losses which can lead to significant economic losses, as well as harm to the reputation of the utility who is operating the power plant. Therefore, it is desirable to detect anomalous behavior in equipment, that in the future, might lead to damage or failure of that or other connected equipment. Equipment damage can increase operations and maintenance costs as well as result in lost revenue for power producers. A robust and sophisticated monitoring program is likely to avoid such incidences. However, due to the heterogeneity of equipment in an electric power plant, it may be difficult and costly to implement a comprehensive monitoring program. A cost-effective approach is to develop a non-invasive, area-wide monitoring system that, from a distance, can collect signals emitted from a variety of electrical and mechanical equipment and perform data analysis that can reveal potential anomalies or provide an early indication of developing malfunctions. Sensing changes in the acoustic and electromagnetic (AEM) signatures of equipment using sensors that monitor sound pressure (acoustic signature) and electromagnetic data (electromagnetic signature) can be incorporated at a particular angle and distance from the equipment to trend changes in equipment performance, and indicate a potential failure. This type of monitoring is simple, cost-effective, and can survey a large area for the purpose of improving equipment reliability.

A survey on robot localization in extraterrestrial environments

Localization plays a key role during autonomous operation of mobile robots in planetary exploration missions. It is a challenging task due to the uncertainty in the environment and limitations in communications. In this paper we discuss various technologies that can be considered for localization of robots in extraterrestrial environments and also analyze various methods that can be implemented using these technologies. After reviewing pros and cons of each method we finally suggest the best technique for localization considering the application of the robot.

Noise analysis methodology for a dual-diaphragm medical device air pump

A noise analysis is conducted on a DC powered dual diaphragm air pump. Using a sound level meter, data acquisition chassis and a LabVIEW based fast Fourier Transform tool, the acoustic spectrum of the pump operating under nominal conditions is obtained. The majority of the acoustic energy propagating from the pump is within the 20 to 2000 Hertz range, classifying the pump acoustics as low pitched and but audible. The large peaks of the power spectrum graphs appear as multiples of the rotational frequency of the DC motor allowing for easy comparison between separate trials. The pump noise is classified as a low pitched with the highest peaks below 60 dB. Inline attenuation devices are placed on the inlet of the pump to neutralize transmitted noise. The effectiveness of the inline filters is assessed using acoustic analysis tools and techniques.

Initial implementation of a novel laser localization system

A novel laser-beacon localization system has been developed which has applications in positioning and navigation of mobile ground or aerial vehicles where other forms of localization are absent (e.g. GPS or radio). The system allows for position determination within an area of interest with reasonable accuracy. Model results and initial prototype have shown that the concept of dual angle-encoded beacons is a valid approach for determining relative location in a sterile environment. A prototype of the system has been realized and initial system testing has been promising with consistent results, indicating that the assumed error levels for the model were reasonable. Testing is underway to validate the results of the model and demonstrate the feasibility of the system. Based on lessons learned thus far, optimizations are planned which should lead to increases in performance of the system as realized in the model, such as actively tracking the location of the beacons once their location is initially known.