Task DRL: Trayectory Tracking

Description

The Robotic Manipulator executes the task of tracking trajectories located within a work zone of the CoppeliaSim simulation scene using a Deep Reinforcement Learning based control approach.

Software:

• CoppeliaSim 4.7 64 bits
• Gymnasium 0.29
• Numpy 1.25
• Matplotlib 3.7.1
• Python 3.11
• Stable Baselines3 2.0

Contenido

• CoppeliaSim Scene for Evaluation and Training.
• .ipynb files for Evaluation and Training.
• Trained model with 10000 episodes (best_model.zip).

Procedure

• The CoppeliaSIm file Entorno_MRR_DRL_IK.ttt is used for training and inference of the DRL model.
• Train: Open, modify parameters of the file TrainingAgentDRL.ipynb as the user sees fit and run training scene in CoppeliaSim.
• Inference: Open the file InferenciaAgenteDRL.ipynb, load the previously trained model and run to evaluate the behavior of the trained agent.

Multimedia content

Youtube link of the model inference: https://www.youtube.com/watch?v=n-YulJUdDHg