Infrastructure-based Vehicle Control
Many situations exist where even the most sensor equipped vehicle will be blind (e.g. occluded dog behind a vehicle). The advantage of infrastructure sensing is that it can fill in missing data and gaps and complement on-board sensing to produce redundancy and an expanded field of view. This project will use a combination of infrastructure sensors (cameras, IR camera, radar, and LiDAR) and deep-learning to build an environment map which will be used by the UNLV CAV to perform safe navigation without any on-board sensing.
Automated Vehicle Simulation
Modern AV design and testing relies heavily on vehicle simulation to create scenarios for repeatability or for safety. Once algorithms have been perfected in simulation they can then be implemented in a real-vehicle with confidence. This project aims to implement a driving simulator on our driving simulator motion platforms. The goal is to connect three different driving simulators (two with three degrees of motion and a desktop-based) for cooperative simulation and testing. The simulation environments under consideration are Gazebo from the Robot Operating System (ROS) or the Carla driving simulator.
Required skills: programming (preferably Python and C++); preferred skills: ROS and driving simulation (e.g. from gaming)
Autopilot Drone for Autonomous Accident Assessment
When traffic accidents occur, it may take hours to clear the accident site using manual methods, and secondary crashes are common during that time. To solve this problem, drones can be used to assess the accident scene in a safer, faster, and more accurate way. A recent study shows that data collection procedures using drones can map a scene in 5-8 minutes. Most existing drones are controlled via short-range signals such as WiFi, Bluetooth, or radio airwave, but an outstanding question is how to control drones from longer ranges. In this project, students will develop an autopilot quaddrone system controlled by 4G cellular network to conduct accident assessment over a 5-mile range. The overall effectiveness of a drone’s operation will be greatly enhanced by the real-time control link, data transfer capabilities, and secure communication offered by cellular networks. The system consists of four subsystems: the autopilot unit, data collector, communication module, and control app. The autopilot unit is built with a Raspberry Pi (RPI), built-in sensors, and a flight controller. The RPI collects inputs from built-in sensors and GPS and sends control signals to the flight controller so that the drone flies along the specified path and collects images at the correct angle. The data collector consists of two high resolution cameras (front and back), GPS, flash light, and other sensors. The communication module can be made with an off-the-shelf extended range and data (XRD) module or 4G LTE dongle. The control app can launch, navigate, and return the drone as well as view the scene in stream. The system can authenticate multiple subscribers – who all receive image data, but only one has control authority. Each subsystem can be developed and tested separately, and the system will be tested on local roads first to evaluate performance in terms of drone balance, command response time, video quality, and power consumption.
Energy Storage as a Service (ESaaS) for Green Mobility
Ideally, one of the characteristics of smart mobility is that it should be green with zero emissions, but the need to charge electric vehicles (EVs) often does not coincide with photovoltaic (PV) power generation due to its intermittent nature. However, battery energy storage, when paired with PV, can provide a range of grid services as well as guaranteed EV charging from 100% renewable energy. While customer-sited battery storage is currently cost-prohibitive, despite significant government incentives, an alternative solution is to lease a portion of the capacity of utility-owned storage systems located at the local substation – a concept referred to as Energy-Storage-as-a-Service (ESaaS). We propose to augment the ESaaS by providing EV owners with guaranteed charging from 100% renewables by dispatching allotted battery capacities for their primary interest (i.e., electricity bill management), and then re-dispatching them to provide multiple, stacked grid services including EV charging. This adds value for both the utility company and the consumer. The optimization of ESaaS is complex due to managing multiple entities with different goals and priorities. We propose to develop algorithms for operating and scheduling an ESaaS using numerical algorithms and Blockchain-enabled smart contracts that will allow PV owners access to additional value streams (i.e., EV charging from renewables) from a utility owned and operated substation battery.
Low-cost Electronically Scanned LiDAR
Sensors and cameras provide mapping and situational awareness for safe navigation of autonomous vehicles. Despite significant advances in this field, many current challenges remain, including the need for reduced-cost, improved reliability sensors that can function under adverse weather conditions. LiDAR provides higher resolution mapping of the environment for autonomous navigation systems, however, a major drawback of current LiDARs is the mechanical scanners used to scan the laser beam, resulting in high cost and large size. A low-cost, compact, electrically-scanned LiDAR can significantly improve safety and reliability of autonomous navigation systems. A promising solution is the use of an electro-optic (EO) scanner to scan the laser beam in X-Y directions; EO scanners are desired due to their compactness, lower cost, higher scanning speed, increased reliability, and potential monolithic integration to laser sources. The objective of this project is to develop a compact electrically scanned LiDAR using embedded quantum dots (QDs). The REU student will be involved in the following tasks: (1) investigation and identification of the QD material for high electro-optic coefficient and optical transparency, (2) investigation and identification of the embedding dielectric for high optical transparency, (3) investigation and determination of optimum QD size and density for high electro-optic coefficient, (4) determination of device layer thickness for optimum EO scanner performance, (5) configuration and design of contacts for the EO scanner, (6) process development for monolithic integration of EO modulator on VCSEL (Vertical Cavity Surface Emitting Laser) chip, (7) fabrication and testing of the monolithically integrated EO scanner on VCSEL, and (8) construction and testing of EO-scanned LiDAR system.