newer version meshcat visualization issue for mesh

[MiaoDragon]

Hello. Thanks for the software effort of the Pinocchio. I've been experiencing a visualization issue in meshcat viewer for meshes. It seems for the new version 3.1 the mesh scale is not correctly displayed when I have a mixed mesh files of STL and DAE. One visualization is shown below for my motoman robot with robotiq gripper, and the gripper has a very large scale. This is the visual geometries. The collision geometries are also problematic.

The scale issue of visual geometrices do not exist for version 2.7, while the collision geometries still have the same issue. The visual geometries are shown as follows:

The visualized model can be extracted from https://github.com/DanManN/motoman

After the model package is installed in a catkin workspace, and the workspace is made, use the following code to visualize the results: (the following code is saved as a file motoman_config.py)

import sys
import os
import rospy
from rospkg import RosPack

class MotomanSDA10F:
    def __init__(self):
        self.ignore_collision_ee_links = [
            "motoman_right_ee",
            "left_outer_knuckle",
            "left_outer_finger",
            "left_inner_finger",
            "left_inner_finger_pad",
            "left_inner_knuckle",
            "right_outer_knuckle",
            "right_outer_finger",
            "right_inner_finger",
            "right_inner_finger_pad",
            "right_inner_knuckle",
            "robotiq_arg2f_extra_link",
            "robotiq_arg2f_base_link",
            "arm_right_link_7_t",
            "arm_right_link_6_b",
        ]
        rp = RosPack()
        self.robot_urdf = rp.get_path(
            'motoman_sda10f_moveit_config'
        ) + '/config/gazebo_motoman_sda10f.urdf'
        package_dirs = []
        for package in ['motoman_sda10f_support', 'robotiq_2f_85_gripper_visualization']:
            path = os.path.join(rp.get_path(package), '..')
            path = os.path.abspath(path)
            package_dirs.append(path)
            # package_dirs.append(rp.get_path(package))
        self.package_dirs = package_dirs

import motoman_config
import pinocchio as pin
import hppfcl
from pinocchio.visualize import MeshcatVisualizer
import time
import numpy as np

motoman = motoman_config.MotomanSDA10F()
model, collision_model, visual_model = pin.buildModelsFromUrdf(motoman.robot_urdf, package_dirs=motoman.package_dirs)
viz = MeshcatVisualizer(model, collision_model, visual_model)
viz.initViewer(open=True)

# Load the robot in the viewer.
viz.loadViewerModel()

# Display a robot configuration.
q0 = pin.neutral(model)
print('q0: ')
print(q0)
time.sleep(1.0)
viz.display(q0)
viz.displayVisuals(True)

The system information is the following:

• OS: Ubuntu 20.04
• Pinocchio version: 3.1.0 and 2.7.0
• meshcat: 0.3.2

Thanks and any help would be appreciated!

[jcarpent]

Thanks @MiaoDragon for reaching us about this bug. I will have a look soon. I can already reproduce your issue.

[jcarpent]

Solved via #2331.