Particle Robotics
The particle robotics project IEEE paper builds on top of simulation work published by our lab with physical particle robots. When these robots of different vibration frequencies are grouped together, they move in predictable xy directions. I worked on the mechanical design & assembly, experiment design, and data processing.
Demo Video
A swarm of particle robots navigating through a narrow corridor.
Mechanical Design
The core requirement of the particle robots is to vibrate. It does this by spinning a mass around the z-axis. During development, I characterized the weight distribution and did tests with different masses and spinning frequencies to find the optimal configuration. Another key feature is the low cost of design and assembly: everything here is laser cut, allowing for faster iteration time and ease of scaling when we started to run larger experiments.
Electrical Packaging
Packaging electrical components was made much easier with a custom PCB. I learned from the EE on our team and helped with getting the boards put together when that became the bottleneck. The loose chain groups together the vibrating particle robots, which causes the emergent motion.
Data Processing
During the data collection and analysis process I optimized the pipeline as much as possible by leveraging CV tools. Using AprilTags to process the video feed and OpenCV to recognize the robots, my pipeline plots the trajectory of the swarm over time. This made data analysis much more efficient with the vast amount of experimentation we did.
