[ English | 한국어 ]

Intelligent Robot Guide

Author: Minsu Kwon (M.S in Dept of AI, University of Seoul)

---

English

Useful docs

• 👉 Install dev packages safely, easily: install_jetson_dev_packages.md
• 👉 Safely setup conda venv for jetson: check_jetson_env_conflicts.md
• 👉 Check current conda venv is conflicted: safe_set_pytorch_conda.md
• 👉 Safely link system packages on conda venv : jetson_system_bridge.md
• 👉 Code examples: code/README.md

Prerequisites

• Hardware: Jetson Xavier NX
• OS: Ubuntu 20.04 (L4T r35.6.x) / JetPack 5.1.5
• Python: 3.8
• Key libs:
  • CUDA 11.4 (JetPack)
  • cuDNN 8.x (JetPack)
  • PyTorch 2.1.0 (Jetson wheel)
  • Torchvision 0.16.1 (from source)
  • OpenCV 4.10.0 (CUDA, installed to /usr/local)
  • (Optional) DeepStream 6.3

Quick Start

1. (Only for UOS class students!) Wi-Fi (AX210) patch + static IPv4

   cd scripts
   bash ./enable_ax210.sh
   bash ./set_wifi_ipv4.sh                 # prompts SSID/PSK, sets static IPv4

---

2. Install jetson dev stack (see install_jetson_dev_packages.md for detail)

   bash ./install_jetson_dev_packages.sh   # default 1,2,3,4,5

3. Safely create PyTorch conda venv for project. (Recommended!)

   bash ./safe_set_pytorch_conda.sh              # choose torch 2.1 / 2.0 / 1.14
   bash ./safe_set_pytorch_conda.sh --no-tv      # if not installing torchvision (time-saving)

4. Check whenever you are doubtful about venv is conflicted - It gives quick solution to resolve the issues

   bash ./check_jetson_env_conflicts.sh

5. Link system packages on current conda venv again just in the case you accidentally installed opencv-python.

   bash ./jetson_system_bridge.sh

---

한국어

참고 문서

• 👉 개발 패키지 안전·간편 설치: install_jetson_dev_packages.md
• 👉 Jetson용 conda 환경 안전 설정: check_jetson_env_conflicts.md
• 👉 현재 conda 환경의 충돌 확인: safe_set_pytorch_conda.md
• 👉 Jetson용 conda에 시스템 패키지를 링크: jetson_system_bridge.md
• 👉 코드 예제: code/README.md

요구 사항

• 하드웨어: Jetson Xavier NX
• OS: Ubuntu 20.04 (L4T r35.6.x) / JetPack 5.1.5
• Python: 3.8
• 핵심 라이브러리:
  • CUDA 11.4 (JetPack 기본 제공)
  • cuDNN 8.x (JetPack)
  • PyTorch 2.1.0 (Jetson 전용 휠)
  • Torchvision 0.16.1 (소스 빌드)
  • OpenCV 4.10.0 (CUDA 지원, /usr/local 설치)
  • (선택) DeepStream 6.3

빠른 시작

1. (서울시립대 수업 전용) AX210 Wi‑Fi 패치 + 고정 IPv4

   cd scripts
   bash ./enable_ax210.sh
   bash ./set_wifi_ipv4.sh                 # SSID/PSK 입력받아 고정 IP 설정

---

2. Jetson 개발 스택 설치 (자세한 내용: install_jetson_dev_packages.md)

   bash ./install_jetson_dev_packages.sh   # 기본 1,2,3,4,5 실행

3. 프로젝트용 PyTorch conda 환경 생성 (권장)

   bash ./safe_set_pytorch_conda.sh              # torch 2.1 / 2.0 / 1.14 선택
   bash ./safe_set_pytorch_conda.sh --no-tv      # torchvision 미설치 (시간 절약)

4. 현재 conda 환경이 충돌을 일으키는지 확인하기 - 문제에 대한 빠른 해결 제공

   bash ./check_jetson_env_conflicts.sh

5. 실수로 opencv-python 등을 설치했을 때, 현재 conda 환경에 시스템 패키지를 다시 링크하기

   bash ./jetson_system_bridge.sh

---