diff --git a/metadrive/tests/test_functionality/test_nondeterminism.py b/metadrive/tests/test_functionality/test_nondeterminism.py
index 95923b75e..57baf4c49 100644
--- a/metadrive/tests/test_functionality/test_nondeterminism.py
+++ b/metadrive/tests/test_functionality/test_nondeterminism.py
@@ -5,8 +5,9 @@
 Usage: run this file. or pytest.
 """
 
+from collections import defaultdict
+
 import numpy as np
-import pandas as pd
 import pytest
 
 from metadrive.envs.metadrive_env import MetaDriveEnv
@@ -42,22 +43,21 @@ def assert_dict_almost_equal(dict1, dict2, tol=1e-3):
 
 
 def are_traces_deterministic(traces) -> bool:
-    df = pd.DataFrame(traces)
-
-    # grouping by repetition to get a list of traces
-    traces = df.groupby("repetition")
-
-    # drop index and repetition ID to compare only step info later
-    stripped_traces = [trace.reset_index(drop=True).drop("repetition", axis=1) for _, trace in traces]
+    # Group traces by repetition
+    grouped_traces = defaultdict(list)
+    for trace in traces:
+        repetition_id = trace["repetition"]
+        grouped_traces[repetition_id].append({k: v for k, v in trace.items() if k != "repetition"})
 
-    # iterate over each trace and check if it is equal to the first one
-    are_equal_to_first_trace = [trace.equals(stripped_traces[0]) for trace in stripped_traces]
+    # Convert traces to lists of dictionaries
+    stripped_traces = [sorted(group, key=lambda x: sorted(x.items())) for group in grouped_traces.values()]
 
-    first_trace = stripped_traces[0].to_dict()
+    # Compare each trace list to the first one
+    first_trace = stripped_traces[0]  # This is a list of dictionaries
     for trace in stripped_traces:
-        # Assert
-        trace = trace.to_dict()
-        assert_dict_almost_equal(first_trace, trace)
+        assert len(trace) == len(first_trace)
+        for i in range(len(first_trace)):
+            assert_dict_almost_equal(first_trace[i], trace[i])
 
 
 @pytest.mark.parametrize(