Filtering by
- All Subjects: Machine Learning
- All Subjects: electric scooter
- Creators: Electrical Engineering Program
Leveraging Machine Learning and Wireless Sensing for Robot Localization - Location Variance Analysis
Modern communication networks heavily depend upon an estimate of the communication channel, which represents the distortions that a transmitted signal takes as it moves towards a receiver. A channel can become quite complicated due to signal reflections, delays, and other undesirable effects and, as a result, varies significantly with each different location. This localization system seeks to take advantage of this distinctness by feeding channel information into a machine learning algorithm, which will be trained to associate channels with their respective locations. A device in need of localization would then only need to calculate a channel estimate and pose it to this algorithm to obtain its location.
As an additional step, the effect of location noise is investigated in this report. Once the localization system described above demonstrates promising results, the team demonstrates that the system is robust to noise on its location labels. In doing so, the team demonstrates that this system could be implemented in a continued learning environment, in which some user agents report their estimated (noisy) location over a wireless communication network, such that the model can be implemented in an environment without extensive data collection prior to release.
The purpose of this project is to create a useful tool for musicians that utilizes the harmonic content of their playing to recommend new, relevant chords to play. This is done by training various Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNNs) on the lead sheets of 100 different jazz standards. A total of 200 unique datasets were produced and tested, resulting in the prediction of nearly 51 million chords. A note-prediction accuracy of 82.1% and a chord-prediction accuracy of 34.5% were achieved across all datasets. Methods of data representation that were rooted in valid music theory frameworks were found to increase the efficacy of harmonic prediction by up to 6%. Optimal LSTM input sizes were also determined for each method of data representation.
The increasing demand for clean energy solutions requires more than just expansion, but also improvements in the efficiency of renewable sources, such as solar. This requires analytics for each panel regarding voltage, current, temperature, and irradiance. This project involves the development of machine learning algorithms along with a data logger for the purpose of photovoltaic (PV) monitoring and control. Machine learning is used for fault classification. Once a fault is detected, the system can change its reconfiguration to minimize the power losses. Accuracy in the fault detection was demonstrated to be at a level over 90% and topology reconfiguration showed to increase power output by as much as 5%.
We present in this paper a method to compare scene classification accuracy of C-band Synthetic aperture radar (SAR) and optical images utilizing both classical and quantum computing algorithms. This REU study uses data from the Sentinel satellite. The dataset contains (i) synthetic aperture radar images collected from the Sentinel-1 satellite and (ii) optical images for the same area as the SAR images collected from the Sentinel-2 satellite. We utilize classical neural networks to classify four classes of images. We then use Quantum Convolutional Neural Networks and deep learning techniques to take advantage of machine learning to help the system train, learn, and identify at a higher classification accuracy. A hybrid Quantum-classical model that is trained on the Sentinel1-2 dataset is proposed, and the performance is then compared against the classical in terms of classification accuracy.
My honors thesis took the form of a creative project. My final deliverables are my research presentation (pdf attachment) and solar powered electric scooter (image attachment). The goal of my project was to fix a second-hand electric scooter and create a solar-powered charger for its battery. The research portion of my creative project focused on exploring the circuit elements in a solar charging schematic and their relationships to power output. First, I explored methods of maximizing power output of the basic solar charging schematic. To find the maximum power output based on different settings of photocurrent (sunlight), I wrote a MATLAB code to calculate maximum power based on its derivative with respect to voltage set equal to zero. Finding this maximum power point in MATLAB allowed me to find its corresponding current and voltage output to produce that exact power. With these max current and voltage values, I was able to solve for an ideal resistor value to set in series with the solar panel in order to achieve these values. In doing so, I designed a maximum power point tracker (MPPT). This became an essential component in my charger’s final design. Next, I explored the microcircuit level of a solar panel schematic. In order to do so, I had to break my single diode model into several diodes in series, resulting in the overall solar panel voltage drop (aka the voltage rating of the solar panel) being divided N times. To find what this N value for a given solar panel is, I performed a lab experiment using a small solar panel and a floodlight to gather the panel’s turn on current and open circuit voltage. These two values helped me find the solar panel’s N value after linearizing the lab data. Now, with a much deeper understanding of solar charging circuitry, I was able to move forward with the design and implementation phase. The design and implementation portion of my creative project included the physical assembly of the solar-powered scooter. First, I analyzed the efficiency differences between having an AC coupled vs. DC coupled system. Due to the added complexity of AC conversions, I deemed it unnecessary to use an inverter in the charger. The charging schematic I designed only called for a charge controller and MPPT, both parts that could easily DC couple the system. Keeping the system in DC from solar panel to battery was definitely the most efficient method, so DC coupling was my final selection. Next, I calculated the required current and voltage output of my charger to meet the specs of the battery and the requirements I set for my project. Finally, I designed a solar array based on these ratings. The final design includes one 30 W panel in parallel with two series-connected 5W panels. The two series panels are affixed on the scooter neck for a built in charge design so that the scooter can be charged anywhere (outside while not in use). The big panel can be connected using a parallel branch in the charging cord that I spliced for added current if charging is set up in a stationary setting (by a window at home). The final design serves the need for sustainable micro mobility in a daily 50% depletion use case kept above 20% charged at all times.