Hello. I'm Trevor Muraro.

The idea of CeilingBot started with a senior project idea that never got off the ground: A robot that could navigate a large, crowded building, such as a hospital or an airport, and do it on a budget. We quickly rejected SLAM as our only means of localization - after all, depending on occupancy conditions a 2-D scan LIDAR might not even be able to see the walls. We also considered RF localization, but discarded the idea due to worries about reflection in an indoors environment.

Our solution was simple: while the robot may not be able to see the walls around it, it can always see the ceiling. Thus, localization is possible with nothing more than an upwards-facing camera and a set of reference markers on the ceiling.

While the project never got off the ground due to the loss of a team member, I am now attempting to implement only the ceiling localization part of the planned robot on a small, cheap platform. This lets me test the functionality of my idea while teaching myself ROS and OpenCV at the same time.

Project AutoLot was a prototype parking management system designed and built by myself and two other students at UC Santa Cruz. Our goal was to help deal with the campus’ parking problems.

This sensor station is my work.  It is capable of monitoring a parking space while communicating with the cloud over secure Wi-Fi and drivers’ phones over Bluetooth, and it can run for over a month on its rechargeable battery.

I designed, developed and tested the entire sensor station - physical construction, electronics and software - with the sole exceptions of the LED placement and enclosure lid. The sensor station is built around a TI CC3220S microcontroller, which provides both control and Wi-Fi functionality. Its main electronic components are mounted on a removable and replaceable “electronics plate” for ease of access and maintenance. The entire system is designed for minimal power consumption, with every active component powered down when not in use.

Not only was my design successful, my attention to detail during hardware and software design and testing enabled me to identify bugs that would otherwise have gone unnoticed.

When I interned at Vigilent in Oakland, I did development and QA for a project based on Node-RED.  When I started, I had no experience with JavaScript, Node.js or Node-RED - and at the time, Node-RED’s test features weren’t documented at all.

In addition to my development and QA work, I researched and compiled documentation on proper module development and testing for Node-RED for the other engineers.  After my internship, I decided to take everything I had learned about JavaScript, Node.js and Node-RED, and write the tutorial I’d like to have had when I started.  This is that tutorial.

Node-RED for JavasSript Newcomers was written for Node-RED version 0.17, in December 2017.

My team designed this robot to solve tasks on a changing course. The catch?  We could only use simple sensors, and we only had five weeks to design the robot in its entirety, build it and test it.

I designed the overall layout of the robot, a seven-sensor tape sensor layout on the robot base that could be used to reliably determine the robot’s precise location on the course, a three-sensor bumper arrangement to align with a very small ball delivery target, and the robot’s ball delivery mechanisms and superstructure. I also designed and wrote the majority of the software developed for the robot, including hardware drivers and multiple state machines.

Not only was GCC-3PO a success, I consider it one of my favorite learning experiences to date.