Open Access. Powered by Scholars. Published by Universities.®
- Keyword
-
- Control (2)
- Neural Networks (2)
- Reinforcement learning (2)
- Autonomous Navigation (1)
- Bot (1)
-
- Competition (1)
- Convolutional (1)
- Convolutional Neural Networks (1)
- Embedded Systems Digital Hardware (1)
- Feedback (1)
- Guided Policy Search (1)
- ILQR (1)
- Image Recognition (1)
- LIDAR (1)
- LQR (1)
- Labeling (1)
- Labelling (1)
- Lidar (1)
- MRAC (1)
- Machine learning (1)
- Navigation (1)
- Optimal Control (1)
- Paths (1)
- Point Cloud (1)
- Poker (1)
- Probabilistic roadmaps (1)
- Rapidly-exploring random trees (1)
- Reinforcement Learning (1)
- Roborodentia (1)
- Robot (1)
- Publication
Articles 1 - 10 of 10
Full-Text Articles in Engineering
Reasoning From Point Clouds, Joey Wilson
Reasoning From Point Clouds, Joey Wilson
Computer Engineering
Over the past two years, 3D object detection has been a major area of focus across industry and academia. This is primarily due to the difficulty of learning data from point clouds. While camera images are fixed size and can therefore be easily trained on using convolution, point clouds are unstructured series of points in three dimensions. Therefore, there is no fixed number of features, or a structure to run convolution on. Instead, researchers have developed many ways of attempting to learn from this data, however there is no clear consensus on what is the best method, as each has …
An Application Of Sliding Mode Control To Model-Based Reinforcement Learning, Aaron Thomas Parisi
An Application Of Sliding Mode Control To Model-Based Reinforcement Learning, Aaron Thomas Parisi
Master's Theses
The state-of-art model-free reinforcement learning algorithms can generate admissible controls for complicated systems with no prior knowledge of the system dynamics, so long as sufficient (oftentimes millions) of samples are available from the environ- ment. On the other hand, model-based reinforcement learning approaches seek to leverage known optimal or robust control to reinforcement learning tasks by mod- elling the system dynamics and applying well established control algorithms to the system model. Sliding-mode controllers are robust to system disturbance and modelling errors, and have been widely used for high-order nonlinear system control. This thesis studies the application of sliding mode control …
Utilizing Trajectory Optimization In The Training Of Neural Network Controllers, Nicholas Kimball
Utilizing Trajectory Optimization In The Training Of Neural Network Controllers, Nicholas Kimball
Master's Theses
Applying reinforcement learning to control systems enables the use of machine learning to develop elegant and efficient control laws. Coupled with the representational power of neural networks, reinforcement learning algorithms can learn complex policies that can be difficult to emulate using traditional control system design approaches. In this thesis, three different model-free reinforcement learning algorithms, including Monte Carlo Control, REINFORCE with baseline, and Guided Policy Search are compared in simulated, continuous action-space environments. The results show that the Guided Policy Search algorithm is able to learn a desired control policy much faster than the other algorithms. In the inverted pendulum …
Weight Controlled Electric Skateboard, Zachary Barram, Carson Bertozzi, Vishnu Dodballapur
Weight Controlled Electric Skateboard, Zachary Barram, Carson Bertozzi, Vishnu Dodballapur
Computer Engineering
Technology and the way that humans interact is becoming more vital and omnipresent with every passing day. However, human interface device designers suffer from the increasingly popular “designed for me or people like me” syndrome. This design philosophy inherently limits accessibility and usability of technology to those like the designer. This places severe limits of usability to those who are not fully able as well as leaves non-traditional human interface devices unexplored. This project set out to explore a previously uncharted human interface device, on an electric skateboard, and compare it send user experience with industry leading human interface devices.
Labeling Paths With Convolutional Neural Networks, Sean Wallace, Kyle Wuerch
Labeling Paths With Convolutional Neural Networks, Sean Wallace, Kyle Wuerch
Computer Engineering
With the increasing development of autonomous vehicles, being able to detect driveable paths in arbitrary environments has become a prevalent problem in multiple industries. This project explores a technique which utilizes a discretized output map that is used to color an image based on the confidence that each block is a driveable path. This was done using a generalized convolutional neural network that was trained on a set of 3000 images taken from the perspective of a robot along with matching masks marking which portion of the image was a driveable path. The techniques used allowed for a labeling accuracy …
The Soul Annoyed Robot: A Senior Project Report, Dayton Andrew Muxlow, Christian Johansen
The Soul Annoyed Robot: A Senior Project Report, Dayton Andrew Muxlow, Christian Johansen
Computer Engineering
Our goal for this senior project was to create a competitive robot designed to com- pete in Roborodentia 2019. Our project started during the Winter 2019 quarter, and ended with the competition on May 18, 2019. During that time, we developed an accurate solenoid shooting mechanism, an elevated conveyor belt to carry poker chips, and a servo arm to scoop in stacks of poker chips. These hardware compo- nents were attached to a circular differential-drive wooden base designed to be easy to control. We also planned out our match strategy and implemented this strategy with software written in C/Wiring to …
Planr.: Planar Learning Autonomous Navigation Robot, Gabrielle S. Santamorena, Daniel Kasman, Jesus Mercado, Ben Klave, Andrew Weisman, Anthony Fortner
Planr.: Planar Learning Autonomous Navigation Robot, Gabrielle S. Santamorena, Daniel Kasman, Jesus Mercado, Ben Klave, Andrew Weisman, Anthony Fortner
Computer Engineering
PLANR is a self-contained robot capable of mapping a space and generating 2D floor plans of a building while identifying objects of interest. It runs Robot Operating System (ROS) and houses four main hardware components. An Arduino Mega board handles the navigation, while an NVIDIA Jetson TX2, holds most of the processing power and runs ROS. An Orbbec Astra Pro stereoscopic camera is used for recognition of doors, windows and outlets and the RPLiDAR A3 laser scanner is able to give depth for wall detection and dimension measurements. The robot is intended to operate autonomously and without constant human monitoring …
Robot Navigation In Cluttered Environments With Deep Reinforcement Learning, Ryan Weideman
Robot Navigation In Cluttered Environments With Deep Reinforcement Learning, Ryan Weideman
Master's Theses
The application of robotics in cluttered and dynamic environments provides a wealth of challenges. This thesis proposes a deep reinforcement learning based system that determines collision free navigation robot velocities directly from a sequence of depth images and a desired direction of travel. The system is designed such that a real robot could be placed in an unmapped, cluttered environment and be able to navigate in a desired direction with no prior knowledge. Deep Q-learning, coupled with the innovations of double Q-learning and dueling Q-networks, is applied. Two modifications of this architecture are presented to incorporate direction heading information that …
Surveying Underwater Shipwrecks With Probabilistic Roadmaps, Amy Jeannette Lewis
Surveying Underwater Shipwrecks With Probabilistic Roadmaps, Amy Jeannette Lewis
Master's Theses
Almost two thirds of the Earth's surface is covered in ocean, and yet, only about 5% of it is mapped. There are an unknown amount of sunken ships, planes, and other artifacts hidden below the sea. Extensive search via boat and a sonar tow fish following a standard lawnmower pattern is used to identify sites of interest. Then, if a site has been determined to potentially be historically significant, the most common next step is a survey by either a human dive team or remotely operated vehicle. These are time consuming, error prone, and potentially dangerous options, but autonomous underwater …
Comparison Of Lqr And Lqr-Mrac For Linear Tractor-Trailer Model, Kevin Richard Gasik
Comparison Of Lqr And Lqr-Mrac For Linear Tractor-Trailer Model, Kevin Richard Gasik
Master's Theses
The United States trucking industry is immense. Employing over three million drivers and traveling to every city in the country. Semi-Trucks travel millions of miles each week and encompass roads that civilians travel on. These vehicles should be safe and allow efficient travel for all. Autonomous vehicles have been discussed in controls for many decades. Now fleets of autonomous vehicles are beginning their integration into society. The ability to create an autonomous system requires domain and system specific knowledge. Approaches to implement a fully autonomous vehicle have been developed using different techniques in control systems such as Kalman Filters, Neural …