Tech Backpack

Infrastructure Resource Kit for Safety AI Support Technicians

Welcome to the Tech Backpack, your go-to reference for the foundational infrastructure knowledge needed to support Safety AI systems in the field. This page is designed to give you a clear overview of the essential networking, operating system, and system-level concepts that form the backbone of our Safety AI products. Think of it as the toolkit you carry in your mind — ready to reference whenever you encounter technical challenges on site.

Inside this Tech Backpack, you will find concise explanations, practical tips, and best practices for managing and troubleshooting the underlying infrastructure. Whether you are connecting cameras, inspecting network configurations, or navigating Linux environments, this guide will help you understand how the system operates and how to maintain its reliability, even before diving into the Safety AI software itself.

What's inside the backpack?

This Tech Backpack is organized into key areas that cover the infrastructure knowledge you’ll need in the field. Each section focuses on practical skills and concepts that directly support the operation, maintenance, and troubleshooting of Safety AI systems. From understanding how devices communicate over a network to navigating Linux environments and verifying hardware connections, the guide ensures you have the essentials at your fingertips.

While the outline may look broad, it is designed to be modular — you can quickly reference the section relevant to the task at hand. Whether you are diagnosing a camera feed, checking system performance, or managing software updates, the Tech Backpack equips you with the foundational knowledge to act efficiently and confidently.

Quick Reference

Here is a quick reference and a cheat sheet for Safety AI related commands.

Troubleshooting

Here's a troubleshooting guide.

Networking

Prerequisite Knowledge

Before diving into networking for Safety AI systems, technicians should be comfortable with:

• Basic computer concepts (what an IP address is, how devices connect to each other)
• Familiarity with operating system navigation (Windows/Linux basics)
• Understanding of wired connections (Ethernet cables, ports)
• Comfort using command-line tools (terminal/CLI)

Checklist

To support Safety AI deployments, technicians need to know how to do the following:

• Verify device IPs and connectivity
• Check network cables and ports
• Set and check IP configuration
• Check internet access
• Check VPN connectivity
• Use these commands: ping, ip addr, nmcli, nmap, ifconfig, ipconfig

Networking Fundamentals

Overview of Safety AI networks

In Safety AI setups, a network is simply how the devices are connected so the system can function.

In practice, this means:

• Cameras send video over the network to the AI device (e.g. Jetson).
• The AI device sends alerts and status over the same network to an IPIO.
• Your laptop uses the network to access, configure, or troubleshoot the system.

For a support technician, the network is:

• The Ethernet cable between devices
• The switch they are plugged into
• The IP addresses that let them “find” each other

If the network is down, the AI might still run — but it won’t see cameras, send alarms, or be reachable for support.

Devices in the network

In a typical Safety AI network, you can expect to see:

• VISBOX, the AI processing unit (e.g. Jetson / on-edge computer)
• Cameras (IP cameras or networked vision sensors)
• Network switch (often PoE and/or industrial)
• IPIO, to connect to external parties over dry contact
• Support laptop (yours, for setup and troubleshooting)
• Remote access gateway / VPN device, probably a Teltonika TRB-140 or RUT200
• Wifi Access Point (AP)

Optional / site-dependent:

• A monitoring PC connected to a display
• PLC / control system (if integrated)
• NVR or video recorder

Physical Layer

Physical layer is the actual hardware and cabling that makes the network work. Most field issues originate here, so this is always the first place to check.

You will typically see:

• Ethernet cables (Cat5e / Cat6)
• Industrial network switch
• Camera network ports
• AI device (Jetson) Ethernet port
• Router / firewall (if present)

First on-site checks:

• Are all cables firmly plugged in?
• Are link lights on at both ends?
• Are there any crushed, bent, or damaged cables?
• Is the network switch powered on?

If this layer fails, nothing above it will work. Make sure you know how to terminate and check a patch cable. Always make sure cables are working when troubleshooting field problems.

Connectivity

Connectivity is how devices in the Safety AI network talk to each other. If devices can’t “see” each other on the network, video feeds won’t reach the AI, alerts won’t reach the IPIO, and troubleshooting is impossible.

Local Area Network (LAN)

• A LAN is the network that connects all devices on-site, typically through a switch.
• All cameras, the AI device, HMI, and sometimes the PLC live on the same LAN.
• Devices on the LAN can communicate directly without going through the internet.

If a Safety AI network includes a GSM gateway or a router connecting the site to the corporate network or the internet, that connection is considered a Wide Area Network (WAN). WAN allows users to connect devices in the LAN area from outside the network, such as over an internet connection. To secure access to the LAN over internet we use a VPN which protects access to the network and hides communication through the internet.

IP Address

An IP address is the unique identifier that allows a device to communicate on the Safety AI network. Safety AI networks use IPv4 addresses, which consist of four numbers (0–255) separated by dots, e.g., 192.168.1.10.

Example IPv4 addresses for a Safety AI site:

• Camera 1: 192.168.1.15
• Camera 2: 192.168.1.16
• VISBOX (AI PC): 192.168.1.90
• IPIO: 192.168.1.10
• Support laptop: 192.168.1.120

These addresses ensure each device can be uniquely identified on the local network. The subnet mask (e.g., 255.255.255.0 or 192.168.1.xxx/24) defines which addresses are on the same LAN, while the gateway defines where traffic goes if reaching outside the LAN.

Static IP

A static IP is a fixed network address manually assigned to a device. In Safety AI networks, every camera, AI device (Jetson), IPIO, and support laptop uses a static IP so that each device can always be reliably reached. Another alternative to distribute IP address is using a DHCP which is a service that leases an IP address to every device that connects to the network.

This method eliminates the need to manually assining IP addresses to each device in the network but makes it impossible to know the address of a specific device. Therefore it is important to know that in Safety AI networks all devices need to be manually assigned an IP address and there are no DHCP services / servers allowed because system communication depends on knowing exactly which IP belongs to which device.

DHCP vs. Static:

• Static IP: manually set, never changes — used for all Safety AI devices.
• DHCP: automatically assigned, can change over time — not used in Safety AI networks, but technicians should understand it to identify misconfigured devices.

Subnet mask and gateway:

• The subnet mask (e.g., 255.255.255.0) defines which devices are on the same LAN.
• The gateway defines where traffic goes if it needs to reach outside the LAN (e.g., via router or GSM gateway).

If there's a GSM router in the network, gateway setting of the Visbox need to show this routers address for Visbox to have internet access.

Connectivity verification

To connect to all devices in the network you need to make sure that you are connected to the network. To check connectivity you need to know your IP address as well IP addresses of other devices.

Knowing your IP address

On a Linux device:

• Open a terminal and run: ip addror ifconfig
• Look for the interface connected to the network (usually eth0 for wired or wlan0 for Wi-Fi).

• The line labeled inet shows the device’s IPv4 address (e.g., 192.168.1.120).

On a Windows device:

• Open Command Prompt and run: ipconfig

• Find the network adapter currently connected. The “IPv4 Address” field shows your IP (e.g., 192.168.1.120).

Make sure the IP is in the same range/subnet as the rest of the Safety AI devices (e.g., 192.168.1.x with subnet 255.255.255.0). If it’s outside the range, the device won’t communicate properly.

Verifying Connectivity (Ping test)

The simplest way to check if another device is reachable is to ping its IP address. You need to know the IP address of the device you want to check connectivity.

Both on Linux and Windows:

• Open a terminal and run: ping 192.168.1.90

How to interpret results

• Reply received: The device is reachable.
• Request timed out / Destination unreachable: The device cannot be reached — check cables, switch ports, or IP configuration.

Tips for technicians

• Always ping cameras and the AI device from your support laptop to verify connectivity.
• If ping fails, try another device on the same switch to isolate whether the issue is the device or the network segment.

Finding devices in the network (nmap)

You can find yourself wandering which devices there are in the network without knowing their IP address. You may want to list all devices in a network you are connected to.

Both on Linux and Windows, use a network scan tool like name (you may need to install it first): nmap -sn 192.168.1.0/24

Accessing the Internet

If you need to access the internet through the internet connection of the system, such as the GSM gateway in the network, make sure your IP settings include a valid DNS address.

DNS must be set in the device’s IP configuration so domain names can be resolved when accessing web interfaces through the system’s GSM gateway. Even when a target IP is provided manually, a valid DNS server is still often required for related services (certificates, redirects, updates) to function correctly. Without DNS defined, access may fail or behave inconsistently.

DNS

DNS (Domain Name System) is the service that translates human-readable names into numeric IP addresses used by networks. In your device’s IP settings, a DNS server must be defined to allow proper name resolution when accessing web interfaces through the GSM gateway. Even if a target is reached via a manually provided IP address, DNS is often still needed for related functions such as certificates, redirects, and background services. Without it, access may fail or behave inconsistently.

Commonly used public DNS addresses

• Google: 8.8.8.8 / 8.8.4.4
• Cloudflare: 1.1.1.1 / 1.0.0.1
• OpenDNS (Cisco): 208.67.222.222 / 208.67.220.220
• Quad9: 9.9.9.9 / 149.112.112.112

Any of these can be set in the device’s DNS field for reliable name resolution.

Configuring DNS on Linux

sudo nmcli connection modify 'Wired connection 1' ipv4.dns "8.8.8.8 8.8.4.4"

Linux

Prerequisite Knowledge

Before diving into networking for Safety AI systems, technicians should be comfortably using a PC with any operating system, such as Windows.

Checklist

To support Safety AI deployments, technicians need to know how to do the following:

• Have remote access to a Linux powered device with SSH
• Copy files with SCP
• Navigate through the file system
• Run console commands

Remote access

For field service techs supporting Safety AI, SSH (Secure Shell) is essential for remote access to Linux systems. SSH allows secure log into a remote Linux device over a network.

Usage

On modern Windows (Windows 10/11), an SSH client is already included. You can use it from PowerShell or Command Prompt:

If it’s missing, you can install it via Optional Features:

• Settings → Apps → Optional Features → Add a feature → OpenSSH Client

To run open a console and type:

• Connect: ssh user@hostname_or_ip such as ssh [email protected]
• Exit: exit

File system

Basic Linux file system operations field tech needs:

• ls – list files
• cd – change directory

• pwd – show current directory
• cp source dest – copy
• mkdir dirname – create directory

Disk usage

Make sure any system would have enough free space on a disk. To check use df -h command.

File transfers

SCP(Secure Copy) is a command-line tool that allows technicians to securely transfer files between local and remote Linux systems over SSH, making it essential for moving logs, configurations, or data during Safety AI support. SCP (Secure Copy) lets you copy files between a local and a remote Linux system over SSH.

scp (Secure Copy) lets you copy files between a local and a remote Linux system over SSH.

Syntax

scp source destination

Examples

• Copy local file to remote: scp localfile.txt user@remote_ip:/remote/path/

• Copy remote file to local: scp user@remote_ip:/remote/path/remotefile.txt /local/path/

• Copy a whole directory: scp -r localdir/ user@remote_ip:/remote/path/

Notes

• Uses the same authentication as SSH.
• Preserves basic permissions but not full metadata unless combined with -p.
• Don’t establish an SSH session first. scp automatically uses SSH to connect to the remote system. You just provide the remote username and host, and it handles the connection, authentication, and file transfer in one step.

System logs

For field technicians, understanding how to read logs is crucial for troubleshooting Safety AI systems. Logs record system, hardware, and application events, allowing techs to quickly identify errors, track service behavior, and diagnose issues without needing to reproduce problems manually.

Reading logs

journalctl – shows systemd logs (system and service messages)

• Example: journalctl -u servicename (logs for a specific service)
• journalctl -f – live follow

Following logs

The command journalctl -u vis -f -b -e does the following:

• -u vis → shows logs for the vis service only.
• -f → follows the log in real time, like tail -f.
• -b → shows logs from the current boot only.
• -e → jumps to the end of the log immediately.

So together, it lets you watch the latest log entries for the vis service in real time, which is useful for troubleshooting Safety AI operations.

Services

systemctl is the Linux command-line tool for managing systemd services, which are background processes that run your applications or system tasks.

Start, stop, restart a service

systemctl start servicename

systemctl stop servicename

systemctl restart servicename

Enable or disable a service at boot

systemctl enable servicename

systemctl disable servicename

List all services

systemctl list-units --type=service

It’s essential for ensuring Safety AI software services are running properly.

Network Management with nmcli

nmcli is a command-line tool for managing network connections on Linux devices using NetworkManager. Field technicians use it to verify, troubleshoot, and configure network settings without a GUI.

Common Commands

• Check network devices and status: nmcli device status

• List all network connections: nmcli connection show

• Connect or disconnect a device: nmcli device connect <device_name> , nmcli device disconnect <device_name>

• Configure a static IP:
  • nmcli connection modify <connection_name> ipv4.addresses <ip>/<mask>
  • nmcli connection up <connection_name>
• Set gateway: nmcli con mod <connection_name> ipv4.gateway 192.168.1.25
• Set DNS: nmcli con mod <connection_name> ipv4.dns "8.8.8.8 8.8.4.4"
• Set IP from DHCP to manual: nmcli con mod <connection_name> ipv4.method manual
• nmcli dev reapply <device> applies the current NetworkManager settings to <the device> without restarting it.

Tips

• Always check device status before troubleshooting connections.
• Use nmcli to quickly switch between networks or apply changes after hardware replacements.

Working with VISBOX

Jetson Orin Systems

Most of the new generation Visbox devices are built on NVidia Jetson Orins. The Jetson Orin is a System on Module (SoM)—it combines CPU, GPU, memory, storage, and I/O interfaces on a single compact board for embedded AI applications. It works on a carrier board, and stores data on an SSD. Visbox devices are powered with external 24V and have a fan-less design which means it relies on its body to cool down the GPU which connects to the body with a thermal plate.

Checking system vitals

On an NVIDIA Jetson Orin, you can check temperature and memory usage using jtop app.

Tools

This section introduces the tools Event Gates develop specifically to support field and support teams in their daily work. It reflects our development team’s ongoing commitment to enabling faster, safer, and more reliable support activities.

Recorder

Recorder is a basic text based tool for recording unprocessed videos from deployed Safety AI systems as well as any connected camera that supports RTSP. Just connect to the Visbox over SSH and type command record. App will guide you through the process.

Downloading recorded videos: scp -r eventgates@VPN_IP:~/Videos . (Replace VPN_IP with the IP address of the Visbox.) To know more about how scp works see File transfers section.

Don't forget deleting recorded videos after downloading. Housekeeping does not track these folders.

Tips for Technicians

1. Do not change IP settings of any device in the network unless you are instructed by Event Gates support team.
2. Do not delete any system files or folder unless you are instructed by Event Gates support team.