At Neuromatch Academy, we hypothesized certain personalities were more distractable, measured by poorer working memory, and proceeded to check via clustering (KMeans, PCA, TSNE) and building a prediction model. Disappointingly, the results were not significant. We'd need to quantify distractability better. Presentation

can't see the pdf? link

A team project for the USS hornet defense hackathon. The drone can take 2D images and calculate the coordinates of these objects in 3D, recognize them and then put them on a map. There’s also celestial navigation for GPS backup.

It was really difficult to work with humanoid robots given numerous servo motor incongruences during calibration. So we threw a hail Mary and just told the robot to go to T-pose on demand, given webcam data. Connect to robot, program fixed points, figure out centre of mass and bounds while playing back pose trajectories.

Repo

Robot lab bench setup, which did not work yet because calibration (to be updated)

Lab bench: