Skip to content

Constrained Hierarchical Deep Reinforcement Learning with Differentiable Formal Specifications

Notifications You must be signed in to change notification settings

a-n-onymous/schrl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Constrained Hierarchical Deep Reinforcement Learning with Differentiable Formal Specifications

Demos

4 robots perform different tasks below. Full demos are collected in a website.

drone-loop.mp4
point-seq.mp4
car-branch.mp4
doggo-signal.mp4

Install

  1. Install this package and dependencies (e.g., pybullet, pytorch, etc.)
pip instal -e .
  1. Follow this document to configure and troubleshoot MuJoCo(-py).

Examples

Differentiable Specifications

We implemented a simple LL parser for our differentiable TLTL specifications.

  1. Syntax Examples: how to write TLTL specifications with pure python built-in operators.
  2. Quantitative Semantics Examples: forward and backward through TLTL specifications.
  3. Specifications: 5 types of specifications and solving them with gradient.

Goal-Conditioned Environments and Policies

Reach Random Goals: An example shows how the goal-conditioned environments and policies work.

Pretrained Policies

Run Pretrained Policies: Run our pretrained policies with

python examples/hrl/pretrained.py \
  --robot_name [point, car, doggo, drone] \
  --task_name [seq, cover, branch, loop, signal] \
  --gui \
  --n_eps [number_of_epochs]

for example, below command runs pretrained policy for point robot's seq task 5 times with gui

python examples/hrl/pretrained.py \
  --robot_name point \
  --task_name seq \
  --gui \
  --n_eps 5

Reuse and Citation

This code repository is packed as a standard python package. Install it with pip install -e ., and one will be able to revoke all the modules. Detailed examples for these modules are provided in examples/.

Tweaking configuration here can change the log level, PyTorch device, etc.

One related paper to this repository is under a double-blind review process. The BibTeX will be available after we can make it public.

About

Constrained Hierarchical Deep Reinforcement Learning with Differentiable Formal Specifications

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages