Arena-Rosnav/rosnav-rl
Modular DRL framework for autonomous robot navigation in ROS2. Plug-and-play RL backends (Stable-Baselines3, DreamerV3), composable reward functions, observation spaces & neural architectures - built for research and deployment.
5 Releases
Latest: 1y ago
Rosnav-RL 2.0v2.0.0Latest
tuananhroman·1y ago·January 26, 2025v1.1.2
ReykCS·3y ago·January 30, 2023Removed robot model from base agent kwargs
v1.1.1
ReykCS·3y ago·January 22, 2023📋 Changes
- Adapted internal naming of encoders.
- Fixed parameter namespace
v1.1.0
ReykCS·3y ago·October 28, 2022In this release we added a new and better way to load models. This is only partly compatible with [Arena Benchmark](https://github.com/Arena-Rosnav/arena-rosnav) and requires specific versions in which the observation space is written to the hyperparameters file.
v1.0.0
ReykCS·3y ago·July 13, 2022📋 Changes
- Provide node for testing and running ROSNav agents in every environment
- Encoders, which provide different observation and action space architectures
- Model with different hidden layer structures