Fast_gicp do not work

[huarzail]

I found that the official code of fase_gicp is linked to the main branch. But when I run the command "bash tum.sh", it was not well running. It endless loop at the second iter(current_pose = self.reg.align(initial_pose)).
The issue suggested I use the branch of gs_gicp_slam.(link: Lab-of-AI-and-Robotics/GS_ICP_SLAM#25)

When I change the fase_gicp brance of gs_gicp_slam, the error changed as : AttributeError: 'pygicp.FastGICP' object has no attribute 'set_max_knn_distance'.

What should I do???

In the fase_gicp brance of gs_gicp_slam environment, I had a test script as:

import pygicp
import numpy as np
from multiprocessing import Process, Queue

def align_task(queue, src_points, tgt_points, initial_pose):
reg = pygicp.FastGICP()

reg.set_input_target(tgt_points)
reg.set_input_source(src_points)

print("initial_guess:",initial_pose)

current_pose = reg.align(initial_pose)

print("current_pose:",current_pose)

queue.put(current_pose)

if name == 'main':
src = np.random.rand(100, 3).astype(np.float32)
tgt = np.random.rand(100, 3).astype(np.float32)
initial_pose = np.eye(4, dtype=np.float32)

queue = Queue()

p = Process(target=align_task, args=(queue, src, tgt, initial_pose))
p.start()
p.join()

if not queue.empty():
    result = queue.get()
    print("Alignment successful. Current pose:", result)

---

Output:

(gsicpslam_01) gvlib_ljh@gvlib-MS-ljh:/3DGS-SLAM/GS_ICP_SLAM/test_ljh$ python test_fast_gicp.py
initial_guess: [[1. 0. 0. 0.]
[0. 1. 0. 0.]
[0. 0. 1. 0.]
[0. 0. 0. 1.]]
(gsicpslam_01) gvlib_ljh@gvlib-MS-ljh:/3DGS-SLAM/GS_ICP_SLAM/test_ljh$

It sees that the code of "current_pose = reg.align(initial_pose)" was not executed.