Autonomous Pet Rescue Robot
Overview
This project involved designing and constructing a fully autonomous robot, as part of a rigorous instrument design course, to compete in the 2025 Engineering Physics Robot Summer Competition at UBC. The competition theme was "Pet Rescue"; the goal of the competition was to then navigate a course along black tape and retrieve 'animal plushies' along the track, to ultimately return them to a safe zone at the start of the course. Some of the pets were obstructed or inside of tight spaces and certain sections of the track are rough and lack the guiding tape. There are many different ways to return the pets to the safe zone, such as driving back, launching them at the basket, and hooking them onto the zip-line, which made the challenge interesting and exciting! Over the course of 2 months, my teammates and I worked tirelessly in the lab to design and build our robot.
Our Design
There are many ways to approach this challenge, but we decided early on that our robot would attempt to collect all of the pets into a basket and then raise this basket onto the zip-line at the end of the course. To better organize the aspects of the design, we split it up between the mechanical design, electrical circuits, and the software/firmware needed to control the robot. Each of my teammates and I took on work from each of these categories and integrated them together to ultimately construct our robot to compete in the race.
Mechanical Design
The robot was as large as possible to accommodate the large basket needed for all of the pets and a cascading lift system was installed on each side of the basket to raise it up high enough to hook into the zip-line. The chassis was made from hardboard cut with the laser cutter supported by rigid aluminum bars which we sized using a water jet cutter. This allowed us to support the large basket as well as the retrieval arm. In order to collect the pets, we designed a robotic arm that works on the principle of a four bar linkage. This allows for the claw at the end of the arm to always be parallel to the ground and removes the need for a third servo motor at the wrist joint found in traditional robotic arms. The claw attached to the end of the arm also used the concept of a four bar linkage to keep the claws straight rather than pinching in. In addition, we used a lazy susan, a pair of gears, and a magnetic angle encoder to allow for the arm to rotate left and right using closed-loop control. Besides learning a lot about designing mechanical devices and using 3D CAD software such as OnShape, I was able to put my machining skills to work utilizing the lathe and water jet cutter.
Electrical Design
Many circuits and sensors were needed to make the robot work autonomously. In order to follow the tape on the surface, we needed IR sensors, and to detect the edge of the track as to not fall off, we used a simple sonar sensor. To detect the pets, we used a time-of-flight array module, which worked as a 8x8 resolution height map. The circuit I worked the most on was the H-bridge motor driver. I started with a breadboard prototype and then designed a printed circuit board using KiCAD. The PCB was able to drive both of the rear wheels as well as the cascading lift mechanism.
Firmware
I programmed an ESP32 micro-controller using the FreeRTOS framework to read sensor inputs, make decisions, and actuate motors. I designed the high-level logic of the robot's control and dug deep into the low-level firmware to efficiently use the limited hardware timers and pwm channels. To implement the logic the robot needs to complete the track and retrieve all of the pets, I decided to use the finite state machine paradigm. Each state has a associated task, and depending on which state it is currently in and what inputs it sees, it may transition to a new state and start executing a new set of instructions. Another important firmware challenge was the tape following. We used infrared reflectance sensors to detect the black tape against the white track, which allowed us to use closed-loop control to accurately follow the tape. We implemented this using a heuristically tuned PID controller. One of my favorite parts of this project was programming the positioning of the arm. Using simple geometry, I was able to translate coordinates in a plane into rotation angles for the servo motors to move to, allowing us to grasp the pets with ease.
Conclusion
After months of spending 10-12 hours a day in the lab and many sleepless nights, our team came away from this project with many valuable lessons in mechanical, electrical, and software design and construction. We also gained profound first-hand insights into prototyping, project management, and systems design. Most importantly, my teammates and I had a blast designing and building this robot and this experience left me with a stronger passion for tinkering and Engineering!
