Collaborative Robotics Final Project
-
Goal: program a LoCoBot to collaborate with other LoCoBots in a resource-gathering task without using digital communication.
-
Each team was only allowed to move certain colored blocks (resources) to stations that each team would decide collaboratively online.
-
My contribution: I used a blob detector to determine the 3D locations of the colored blocks, to be used in the motion planner.
DEEP LEARNING Final Project
-
In this project, we explored several techniques to improve upon a network architecture that generates a 3D point clouds from 2D images.
-
The improvements include leaky ReLU and dropout, inception with dimension reduction, multi-head attention, category information integration, and description information integration.
Decision Making Under Uncertainty Final Project
-
My team and I used Monte Carlo Tree Search to determine a winning policy in the game Banananagrams.
-
I wrote the functions to determine the possible actions given the current state, and I conducted experiments to determine the optimal depth and number of simulations.
-
Our optimal policy performed comparably to a novice human player.
-
See the final report here.
PRINCIPLES OF ROBOT AUTONOMY FINAL PROJECT
-
My team and I implemented controls, motion planning, perception, localization, and decision making on a TurtleBot using ROS2.
-
We enabled the TurtleBot to explore and map its environment and to stop moving when it sees a stop sign.
mechatronics Final Project
As part of the Stanford course ME 218A: Smart Product Design Fundamentals, I worked with a team of two other graduate students to create a mechatronic machine, the Yogi PAL. This machine helps airline fliers relax by guiding the user through three calming interactions. More information about this project can be found here.
