This project is used to embed the ltl tasks. This repo is built using bebbo-LTL2Action and LTL2Action

The repository implements training the embedding of LTL tasks through reinforcement learning executed on the LTLbootcamp environment. More details can be found in the Report.

• Stable-Baselines3
• gym-minigrid

or, for better compatibility, the following command can be used:

conda env create -f environment.yml

python pretrain.py

• Vaezipoor, Pashootan, Li, Andrew, Icarte, Rodrigo Toro, and McIlraith, Sheila (2021). “LTL2Action:Generalizing LTL Instructions for Multi-Task RL”. In:International Conference on MachineLearning (ICML)
• Icarte, Rodrigo Toro, Klassen, Toryn Q., Valenzano, Richard Anthony, and McIlraith, Sheila A.(2018). “Teaching Multiple Tasks to an RL Agent using LTL.” In:International Conferenceon Autonomous Agents and Multiagent. Systems (AAMAS), (pp. 452–461)

• Write code to evaluate the similarity among ltl tasks and evaluate the learnt embedding model.
• Construct the LLM model using langchain.
  • 1. Construct the model to generate similar ltl tasks and corresponding behaviors, and save them into possbible errors-revised task pairs. (done)
  • 2. Construct the embedding and use the retrieved content to construct the prompt. (done)
  • 3. Construct groundtruth dataset.
  • 4. How to build up robotic experiments to evaluate?