PEDRO A. PEÑA
GENERATIVE AI
Researching Generative AI to develop novel 3D graphics technologies. Researched transformers, diffusion, CLIP, VQA, and other novel models to develop 3D applications such as generative character animation.
ROBOTICS AND AUGMENTED REALITY
Using the Magic Leap One device to develop a human-robot interface that enables humans to interact with a virtual robot utilizing augmented reality. The interface uses ROS# to communicate with a ROS server and enables a user to map an environment, place and navigate a robot, and label the environment with markers that can be sent to ROS. This project is part of a Magic Leap grant.
PEDESTRIAN FORECAST MODEL
In this research, I introduce inverse trajectory planning for robotics systems to forecast the pose of a human by fusing sensor data with a probabilistic model. I validate the model with Toyota Human Support Robot and Waymo Open Dataset.
HUMANOID ROBOTICS
Motion planning for kick trajectories while a humanoid robot is walking. I also worked on a ZMP controller for balancing while generating kick trajectories using cubic splines, sextic splines, and cubic Hermite splines.
T-TR1
Working on a telepresence robot that will enable humans to interact with other humans in different parts of the world at the Tokyo Olympics 2020.
T-HSR
A robot platform for human-robot interaction. I worked on grasp planning, navigation, behavior planning, human-robot interfaces, and forecasting human motions.
FOLLOW ME!
We used ROS2 as a platform to conduct AI research for developing a Follow-Me capability as a proof- of-concept on a wheeled robot, demonstrating that AI research is possible in the ROS2 framework.
IR BEACON TRACKER
A nonlinear optimization algorithm to find 3D features from 2D features. The application was following a person where a person wears an infrared transmitter.