[BUG] Sokoban envs crash consistently

[leor-c]

🐛 Bug

The Sokoban environment crashes after a while when calling step. I've reproduced this behavior with MiniHack-Sokoban1a-v0, MiniHack-Sokoban1b-v0 and MiniHack-Sokoban2a-v0 (didn't test other versions but it looks consistent).

To Reproduce

Steps to reproduce the behavior:

1. Reset the environment.
2. Take a sequence of random actions, at some point the step function will crash (prior to episode end).

Code:

import gym
import minihack
import numpy as np

env = gym.make("MiniHack-Sokoban1b-v0")
num_actions = env.action_space.n
env.reset()
done = False

while not done:
  env.step(np.random.randint(num_actions))

Trace:

Traceback (most recent call last):
  File "<stdin>", line 2, in <module>
  File "/root/miniconda3/lib/python3.10/site-packages/minihack/envs/sokoban.py", line 30, in step
    return super().step(action)
  File "/root/miniconda3/lib/python3.10/site-packages/minihack/base.py", line 396, in step
    return super().step(action)
  File "/root/miniconda3/lib/python3.10/site-packages/nle/env/base.py", line 373, in step
    end_status = self._is_episode_end(observation)
  File "/root/miniconda3/lib/python3.10/site-packages/minihack/envs/sokoban.py", line 41, in _is_episode_end
    agent_pos = list(self._object_positions(observation, "@"))[0]
IndexError: list index out of range

Expected behavior

should not crash.

Environment

MiniHack version: 0.1.6
NLE version: 0.9.0
Gym version: 0.22.0
PyTorch version: 2.3.1
Is debug build: No
CUDA used to build PyTorch: 12.1

OS: Ubuntu 22.04.4 LTS
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
CMake version: version 3.22.1

Python version: 3.10
Is CUDA available: Yes
CUDA runtime version: Could not collect
GPU models and configuration: GPU 0: NVIDIA GeForce RTX 4090
Nvidia driver version: 555.42.06
cuDNN version: Could not collect

Versions of relevant libraries:
[pip3] numpy==1.24.2
[pip3] torch==2.3.1
[pip3] torchaudio==2.3.1
[pip3] torchvision==0.18.1
[conda] blas 1.0 mkl
[conda] ffmpeg 4.3 hf484d3e_0 pytorch
[conda] libjpeg-turbo 2.0.0 h9bf148f_0 pytorch
[conda] mkl 2023.1.0 h213fc3f_46344
[conda] mkl-service 2.4.0 py310h5eee18b_1
[conda] mkl_fft 1.3.8 py310h5eee18b_0
[conda] mkl_random 1.2.4 py310hdb19cb5_0
[conda] pytorch 2.3.1 py3.10_cuda12.1_cudnn8.9.2_0 pytorch
[conda] pytorch-cuda 12.1 ha16c6d3_5 pytorch
[conda] pytorch-mutex 1.0 cuda pytorch
[conda] torchaudio 2.3.1 py310_cu121 pytorch
[conda] torchtriton 2.3.1 py310 pytorch
[conda] torchvision 0.18.1 py310_cu121 pytorch

Additional context