Final Weeks. Final part of the work
The report and presentation is written and a final video is created
Logbook
2020
Weeks 52-54. Training of the reinforcement learning algorithm
Combination of parameters for the different trainings
Weeks 48-51. Preparing the different trainings
The parameters that will be changed in the different trainings are fixed
Weeks 44-47. Summer advance. Training using qlearn with the camera
Advances with the formula 1 training using the camera with the qlearn algorithm.
Week 36-43. F1 complete a lap
The Qlearning algorithm manages to complete one lap of the circuit.
Week 35. New trainings with jumps
Try new training with ‘jump’ mode. Every reset is in a different place
Week 31-34. New trainings
New tests changing the possibilities of the environment
Week 30. First training
Finishing the creation of the formula 1 exercise and ready for testing.
Week 28-29. Refactoring and improving the F1 environment
Finishing the creation of the formula 1 exercise and ready for testing.
Week 26-27. Migration from the Turtlebot example to the Formula 1
Going deeper into the steps performed by the algorithm.
Week 25. Deep in into code
Going deeper into the steps performed by the algorithm.
Week 24. Turtlebot + Camera
Running and training the turtlebot example using camera.
Week 23. First steps on Gym-Gazebo
First steps with Gym-Gazebo Installing and running examples.
Week 22. Change to Gym-Gazebo
Change of course. Change from Gym-Pyxies to Gym-Gazebo as intermediary with Gazebo.
Week 21. Deep into Gym-Pyxis.
Going deeper into the gym-pyxis library and trying to create a simple client.
Week 20. Networking and learning in the Gym-Pyxies tool.
Training of a network given a dataset and learning in the tool Gym-Pyxies for a first approach between Gazebo and OpenAI-gym.
2019
Week 19. Testing and exploring the pilot code.
New structure of Neural Behaviors repository and deeping the Vanessa’s code.
Week 17-18. Continue with DQN algorithm. Problems with graphic card (Part II).
Finishing the DQN algorithm for the game of pong. Exploring the pilot code to implement the equivalent in learning by reinforcement.
Week 15-16. Continue with DQN algorithm. Problems with graphic card.
Solving problems with the graphic card and reading more information about the DQN algorithm.
Week 14. Organization and first algorithm using Deep Learning: DQN.
Execution of the Pong game from the code provided in Alberto’s repository.
Week 13. Installing nvidia-docker and undestanding DRL algorithms (Part II).
Studying the methods seen the previous week (continuing).
Week 12. Installing nvidia-docker and undestanding DRL algorithms.
Studying the methods seen the previous week.
Week 10-11. Running different exercises from repo
Running different methods and agents for the resolution of DRL problems.
Week 9. Landing on the working infrastructure
Doing some examples and installing infrastructure.
Week 7-8. Reading more documentation and small tutorials.
Apply what you have learned to a small tutorial
Week 6. Summary of definitions and concepts.
Definitions of agents, environments, states, actions, rewards, …
Week 1-5. Previous Work. Soaking me in knowledge.
Reading articles, tutorials and collecting information
Week 0. Hello World
Introduce to myself