instadeepai / jumanji

🕹️ A diverse suite of scalable reinforcement learning environments in JAX
https://instadeepai.github.io/jumanji
Apache License 2.0
613 stars 78 forks source link

Fix: robot warehouse rewarding non-goal states #190

Closed arnupretorius closed 1 year ago

arnupretorius commented 1 year ago

The logic for checking whether a shelf is at a goal state upon delivery before providing a reward has a bug (i.e. shelves are being rewarded at none goal states). The suggested fix is to decompose the conditional (as was originally done).