Releases: MIT-TESSE/tesse-gym
Releases · MIT-TESSE/tesse-gym
0.2.0
Adds features for the ICRA 2022 paper Hierarchical Representations and Explicit Memory: Learning Effective Navigation Policies on 3D Scene Graphs using Graph Neural Networks.
0.1.6
Changes
- Update some default params
- Update how we sync with perception pipeline
0.1.5
Provides ability to integrate with a perception pipeline such as kimera.
0.1.4
Add error handling for network timeout issues.
0.1.3
Changes:
- Update baseline PPO default params
- Update tesse-gym params
- Add continuous controller collision check that doesn't rely on metadata
- Clean up continuous controller (use namedtuples instead of dictionaries)