Yu Peng

Real-Time OS

---

In a team of 2, we built a simulated Auto Braking System for cars. This involved using the FreeRTOS library on an Arduino Uno to ensure deadlines were met. For example, automated braking had to happen as fast as possible, while printing to the UART terminal should happen at least once every second.

The most challenging aspect of this project was when we ran out of SRAM on the Arduino Uno, which was a mere 2 KB. That was when things started to malfunction with no apparent reason. In the end, we had to optimize the use of FreeRTOS data structures, such as queues and semaphores, to make it all fit.

• Type: Academic
• Date: April 2017
• Language: C

Close

Embedded Programming

---

In a team of 2, we programmed the LPC17xx, attached to a baseboard, to care for the elderly. We interfaced with various peripherals: light sensor, accelerometer, temperature sensor, OLED screen to display sensor readings and so on.

Although it was tedious to check through different documentations just to get something to work, I really enjoyed working with such low-level applications as I had full control of the hardware. Working on a development board also meant we could change the wiring of pins to enable a previously unavailable function.

The most challenging aspect was debugging because there could be mistakes anywhere from the hardware configuration to the software logic (or even both!). We had to be very methodical in isolating the problems.

• Type: Academic
• Date: April 2017
• Language: C

Close

Python Web Crawler

---

I wanted to pick up a scripting language, something different from what I already knew. Since the best way to learn is through practice, that is exactly what I did. The awesome web crawler tutorial by thenewboston formed the basis of my script. I then modified it to scrape and download recent soccer highlight videos.

The most challenging aspect was the transition from strongly-typed languages as well as getting used to features and data structures that are unique to Python.

• Type: Personal
• Date: Dec 2016
• Language: Python

Close

Software Engineering

---

In a team of 4, we wrote a To-Do-List desktop application for Windows. Our implemented features included: CRUD, undo and redo, search and so on. We applied software engineering principles to ensure modularity and ease of understanding of our code. We also wrote JUnit tests to ensure its correctness.

The most challenging aspect was in the initial phase where we had to decide on the software architecture, since that had far-reaching implications. Yet, after countless hours at the whiteboard, there were still tweaks to be made when the actual coding began.

• Type: Academic
• Date: Nov 2016
• Language: Java
• Github

Close

Field-Programmable Gate Array Programming

---

The Basys3 board was programmed to generate various signals at specified frequencies and amplitudes: sine wave, triangle wave, square wave and so on. A keyboard using the PS/2 protocol was also interfaced with the FGPA board to specify precise frequencies and amplitudes.

The most challenging aspect was that it takes perhaps 10 minutes to compile the code and program the FPGA board. It was very frustrating for it to have gone through half the build and then spew a syntax error.

• Type: Academic
• Date: Nov 2016
• Language: Verilog

Close

Android Social Alarm Clock

---

Together with a friend, we built an alarm clock application that lets a user request for another user to wake them up. At the arranged time, the 'waker' can send voice messages (in assorted accents) to wake the other person up.

The most challenging aspect was getting regression bugs due to the lack of unit testing. Neither of the terms 'regression' and 'unit testing' made sense to us. The result: manually going through the lines of code and adding print statements to find the error.

• Type: Personal
• Date: Aug 2016
• Language: Java

Close

Arduino Line-Follower

---

In a team of 3, we assembled the circuit and wrote the logic for our 4-wheeled vehicle to travel along a black line. Detection of the black line was done using light-dependent resistors.

The most challenging aspect was the inconsistent behaviour that had arisen due to the breadboard. This was resolved when we transferred everything to a protoboard and soldered the components down.

• Type: Academic
• Date: May 2016
• Language: C

Close