Solve a collection of Data Science projects using Artificial Neural Networks based on projects from Udacity, Coursera, and Kaggle GitHub
Topics: data processing, visualization, DNN/CNN/RNN Modeling, NLP
Resources: Jupyter, Pandas, Numpy, Seaborn, TensorFlow, Keras, and Scikit-learn
RL-ROBOT: Reinforcement Learning framework for Robotics
2015 - 2016
Implemented a practical Reinforcement Learning framework from scratch for easy experimentation with simulated and real robots in V-REP and ROS GitHub
Applied a value-iteration-based method to several robotic tasks with a novel algorithm that improves softmax action selection by structuring the inputs
The resulted Reinforcement Learning method enhances model-free solutions with minimal task knowledge
It is also compatible with high-dimensional Reinforcement Learning with low computational cost
Contributions to Remote Optimal, Adaptive Control of Mobile Robots
2012 - 2015
Developed a full-stack system with a novel graphical interface and sensors-actuators (real and simulated) modules for controlling mobile robots Project Log
Includes a hierarchical regulator of sensor data transmission for networked telerobots
Also includes a functional collaborative navigation system for path following in real and simulated robots
Teaching Reinforcement Learning with Lego Mindstorms NXT
2012 - 2015
Studied the viability of Reinforcement Learning methods in small robots Project Log
Designed a minimal framework that adapts value-based Reinforcement Learning algorithms in a Mindstorms NXT mobile robot
Focused on applicability and flexibility to perform several academic tasks
Later used in mini-projects for MSc students of cognitive robotics, thus improving the teaching experience