Skip to content

Latest commit

 

History

History
62 lines (36 loc) · 1.85 KB

installation_and_troubleshoot.md

File metadata and controls

62 lines (36 loc) · 1.85 KB

Installation Steps:

  1. Install Miniconda:

    • Download Miniconda from the official website and follow the installation instructions for your operating system.
  2. Create a Virtual Environment:

    conda create -n quadrobo_env python=3.12
    conda activate quadrobo_env
  3. Install Required Packages:

    pip install gymnasium[mujoco] matplotlib hydra-core
  4. Install CadQuery:

    pip install cadquery
  5. Install quadrobo_gym Custom Environment:

    If quadrobo_gym is not available via pip, ensure you have access to its source code and install it using:

    pip install /path/to/quadrobo_gym

Note: Always ensure that your package versions are compatible to prevent such errors. Regularly check the official repositories and documentation for updates and fixes related to these issues.

Troubleshooting Known Issues:

To address the issues you've encountered, follow these troubleshooting steps:

1. AttributeError: module 'numpy' has no attribute 'bool8'

This error arises due to incompatibility between CadQuery and numpy version 2.0.0, where the bool8 attribute has been removed. To resolve this:

  • Downgrade numpy to a compatible version:

    pip install numpy==1.26.4

This solution is discussed in CadQuery's GitHub issue #1626.

2. AttributeError: 'MjData' object has no attribute 'solver_iter'

This error occurs because, in MuJoCo version 3.0.0, the solver_iter attribute has been renamed to solver_niter. To fix this:

  • Update your code to use the new attribute name:

    Replace instances of solver_iter with solver_niter in your codebase.

This change is detailed in Gymnasium's GitHub pull request #746.