EECS 149
Contents
Home
Overview
Logistics
Technology
Schedule
Reading
Assignments
Lab & Project
bCourses
Piazza
Reading
References
Resources
Course Development
Wiki
SVN
|
Suggested Project Topics
Below we give a list of possible project topics for EECS 149, with potential mentors also listed alongside. You are also encouraged
to formulate your own project topics. Note that each project must directly relate to at least two distinct topics covered in the lectures.
- Semi-Autonomous Driving (Mentor: Dorsa Sadigh)
This project will imitate shared human-autonomous control of cars, similar to the advanced driver assistance ("self-driving") features arriving in today's cars.
We will simulate a semiautonomous driving scenario using small toy cars (similar to RC cars) mounted by lidars and other sensors.
We consider city driving scenarios, where an autonomous controller only exist under a set of environment assumptions.
We propose monitoring environment assumptions in real time, and determining when to transfer control to/from the human driver in a timely manner.
- Accessors for the Cyber-Physical Swarm (Mentors: Matt Weber, Chris Shaver, Marten Lohstroh)
Accessors are an active research project in the TerraSwarm group. An Accessor is a local proxy for a remote services that facilitates composition of
services in the coming Swarm of ubiquitous connected devices. Click here
for a reference about accessors. The following project topics involve creating and deploying accessors in various applications of interest:
- Kobuki Accessor:
Drive a Kobuki over the Internet. The myRIO you have been using in lab to control the Kobuki robots has WiFi capability.
Write an accessor that sends commands and receives data from the Kobuki over the network.
-
IoT Device Accessor:
This project applies to just about any hardware device you want to build or buy with network capabilities.
Just like the Kobuki Accessor project, write an accessor for it that sends commands and receives data over
a network (doesn't have to be WiFi).
- MPC on Embedded Platforms (Mentor: Vasu Raman, UTC)
MPC, or Model-Predictive Control, is an optimization-based approach to
control where a model of the plant being controlled is used. This project focuses
on iterative, finite horizon, discrete time optimization of a model of the
plant. The idea is to implement real-time MPC on embedded platforms and verify
that it achieves the control objectives within real time bounds. For example,
consider implementing an inverted pendulum mounted on top of a hill climbing Kobuki,
where MPC is used for the inverted pendulum.
- Real-Time CyberSim/CPSGrader (Mentors: Alex Donze, Trung Tran)
In the labs you used CyberSim with an auto-grader that operates on simulation traces.
Modify the auto-grader back-end of CyberSim, known as CPSGrader, to operate on
real-time traces obtained from a robot (e.g. the Kobuki) via WiFi.
This will implement a real-time monitoring and debugging engine for the Kobuki.
- Controlling wearable LEDs (Mentors: Hugo Andrade and Trung Tran, NI)
Create a wearable gadget with a controllable LED display that varies based on
gestures, body movement, or other human-provided control.
- Quadrotor based myRIO controller (Mentors: Hugo Andrade and Trung Tran, NI)
myRIO-controlled quadcopter, see for example these videos:
Video 1 and
Video 2.
- Gesture-controlled music (Mentors: Hugo Andrade and Trung Tran, NI)
See, for example, Guitar playing:
Video 1 and
Video 2.
- Investigating vulnerabilities related to MEMS Sensors (Mentor: Yasser Shoukry)
MEMS-based sensors (e.g. accelerometers and gyroscopes) are widely
used in embedded systems. However, MEMS sensors are
sensitive to both electromagnetic and acoustic interferences
which can be used to launch adversarial attacks. The aim of
this project is to study this attack surface experimentally to
understand the severity of such attack surface.
See Chapter 17 of Lee & Seshia for some background.
- Privacy implications of WiFi Backscattering waves: (Mentor:
Yasser Shoukry)
Backscattering is a physical phenomena were EM waves reflects back
from moving bodies. Recently this phenomena has been used to
detect human gestures by relying on the backscattering from
WiFi signals that exist in nearly every house. However, this
phenomena opens the back door to build adversarial devices
that leak information about house residents. In particular,
this project aims to modify any existing WiFi router in order
to use backscattering to continuously leak the following kinds of
information about house residents: 1) Is there anyone
currently at house? 2) How many residents are currently inside
the house? and 3) who exactly are inside the house?
A few other project examples, ideas, and links:
- H2O_IQ uses a moisture sensor for monitoring plants, a Raspberry Pi as the computing unit, an XBee radio to communicate, and a servo actuator to control irrigation.
- SmartThings offers a set of sensors and actuators so that people can easily build their own automated smart home.
|