Effective Synthesis of Robotics Programs from Demonstrations and Safety/Reachability Specifications
-
- Go the root directory:
cd roboSpec
- Go the root directory:
-
- Run
git submodule update --init --recursive
- Run
-
cd pips
-
git checkout base
-
- Run
make
- Run
-
- Go the root directory:
cd roboSpec
- Go the root directory:
-
- Run
./scripts/init.sh
- Run
-
- Run
./scripts/simul_gt.sh
to generate the initial samples from the ground truth policy.
- Run
-
- Run
./scripts/run_pips.sh
to run pips on the generated samples and write (Pythonic version of) the learned policy inlearned_policy.py
file.
- Run
-
- Run
python robo_spec.py ldips
. The behaviors of this policy will be plotted inplots/policy_ldips/distance.png
.
- Run
-
- Append the repaired samples from the above simulation to the sample file
python scripts/append_repaired_samples.py
- Append the repaired samples from the above simulation to the sample file
-
Repeat from step 4 to learn a new policy using the repaired samples.