Blog

Wi-Fi, LoRa or LTE: Choosing Comms for an Outdoor Robot

Tom Cosgrove,

Reading Room: The Fictionlab Blog

TL;DR: For outdoor robot communication, Wi-Fi gives you high bandwidth at short range (typically 50-150 m line-of-sight in the field), LTE offers wide-area coverage with 30-150 ms latency but depends on carrier reception, and LoRa provides multi-kilometer reach at kilobit-per-second speeds. Most field deployments combine two channels: LTE or Wi-Fi for teleoperation and video, LoRa for backup telemetry and low-rate contingency messages. Choose based on range, payload, latency budget, and whether you can rely on infrastructure. Safety stopping must be handled locally and must not depend solely on any radio link.

Why outdoor robot communication is harder than it looks

Indoor robots operate in a controlled RF environment. Outdoors, you deal with terrain occlusion, weather attenuation, multipath from vegetation and water, and interference from other radios sharing the same ISM bands. A link that shows 100 Mbps on a rooftop test will drop to 5 Mbps when the robot rolls behind a building or into a ditch.

Before picking a radio, define three numbers: the maximum operating range in meters, the required uplink and downlink bandwidth in Mbps, and the tolerable round-trip latency in milliseconds. Add a fourth if you run ROS 2: whether you need reliable QoS for critical topics like /cmd_vel. Emergency-stop and other safety functions should be implemented independently of ROS 2 QoS and radio connectivity. These four parameters eliminate most options immediately.

The Leo Rover ships with a 2.4/5 GHz Wi-Fi module by default, and many integrators add LTE or LoRa modules for specific missions. The choice is not academic. It defines what your robot can do when it leaves the lab.

How Wi-Fi performs for field robots

Wi-Fi (802.11n/ac/ax) is the default for short-range teleoperation because it handles ROS 2 DDS traffic well, supports multi-megabit video streams, and adds only a few milliseconds of latency on a clean link.

Real-world outdoor range depends heavily on antenna gain and line-of-sight. Standard omnidirectional setups at 2.4 GHz typically give you 80-150 m of usable range in open terrain, dropping to 30-50 m through light vegetation. Point-to-point directional links (like a Ubiquiti Bullet or Rocket paired with a sector antenna at the base station) can extend this to 1-2 km, but require line-of-sight and appropriate antenna coverage as the robot moves.

The 5 GHz band gives you more channels and less interference, but attenuates faster through obstacles. The 2.4 GHz band travels farther but is crowded almost everywhere near people.

Use Wi-Fi when:

  • Your mission stays within a fenced site, farm plot, or construction area under 200 m across.
  • You need HD video, LiDAR streams, or full ROS 2 topic bridging.
  • You can install a base station or mesh nodes on the site.

When LTE (and 5G) makes sense

LTE turns locations with cellular coverage into potential control sites. You gain wide-area range at the cost of latency, dependency on a carrier, and monthly data fees.

Typical LTE round-trip latency to a public cloud VPN endpoint sits between 30 and 150 ms, depending on carrier load and signal strength. Uplink bandwidth (robot to operator) is usually the bottleneck: expect 5-20 Mbps in good coverage, sometimes under 1 Mbps at cell edges. Downlink is comfortable for command traffic.

For teleoperation over LTE, tunnel ROS 2 traffic through a VPN (WireGuard is popular for its low overhead) or use a dedicated teleoperation stack that handles NAT traversal and adaptive video bitrate. Raw DDS discovery does not reliably traverse carrier NAT, and multicast discovery should not be assumed to work across routed VPN connections.

Consider LTE when:

  • The robot operates over kilometers or moves between disconnected sites (typical for inspection missions across pipelines, solar farms, or agricultural fields).
  • You cannot deploy your own base station infrastructure.
  • Occasional 100-200 ms latency spikes are acceptable for your control loop.

5G can provide lower latency and higher throughput than LTE, especially on well-configured standalone networks, but real-world performance and coverage vary widely, particularly outside densely populated areas. Private 5G networks are becoming viable for industrial sites, though setup cost is significant.

What LoRa is actually good for

LoRa is not a teleoperation link. It is a long-range, low-power telemetry channel that trades bandwidth for range and battery life. A LoRa module at 868 MHz (EU) or 915 MHz (US) with a standard antenna reaches 2-5 km in open terrain and can exceed 10 km with elevated antennas and clear line-of-sight.

The catch: effective throughput is measured in bits per second, not megabits. At spreading factor 7 you get around 5 kbps on a typical 125 kHz channel; at SF12 you drop to around 300 bps. Regional duty-cycle, dwell-time, power, and channel-use rules can further limit how often you can transmit.

What you can send over LoRa:

  • GPS position updates every few seconds.
  • Battery voltage, motor current, temperature.
  • Simple commands: change mode or request a return-home behavior.
  • Low-rate telemetry when the primary link drops.

What you cannot send: video, LiDAR scans, ROS 2 topic streams, or anything approaching real-time control feedback. Do not rely on LoRa as the sole emergency-stop path; the robot needs an onboard fail-safe that stops or enters a safe state when communications are lost.

How to compare Wi-Fi, LTE, and LoRa side by side

The following table summarizes typical field values. Treat them as starting points, not guarantees. Actual performance depends on antennas, terrain, and interference.

Parameter Wi-Fi (2.4/5 GHz) LTE / 4G LoRa (868/915 MHz)
Realistic outdoor range 50-200 m (up to 1-2 km directional) Wherever carrier reaches 2-5 km typical, 10+ km LOS
Throughput 10-100+ Mbps 1-20 Mbps uplink 0.3-22 kbps on typical 125 kHz channels
Latency (RTT) 2-10 ms 30-150 ms Hundreds of ms to seconds
Power draw (radio) 1-3 W 2-5 W (peaks higher) Tens to hundreds of mW while transmitting; much less in sleep
Hardware cost Low Medium (module + SIM) Low
Recurring cost None Data plan per SIM None (or LoRaWAN gateway fee)
Formal requirements License-free, respect channel/power limits SIM, carrier contract, IMEI registration in some countries License-free ISM, regional duty-cycle, dwell-time, and power rules apply

When to combine two channels

Single-link designs can fail badly in the field. A robot that loses its only radio can become an expensive paperweight or, worse, unsafe if it lacks an onboard fail-safe. Dual-radio architectures are common for serious outdoor deployments.

Common combinations:

  • LTE for video and control, LoRa for emergency telemetry. If the LTE modem loses signal or the SIM gets throttled, LoRa can still report position and carry low-rate contingency messages. The robot must independently enter a safe state if its primary control link is lost. This is a common setup for kilometer-scale outdoor robot communication.
  • Wi-Fi on site, LTE as fallback. The robot uses local Wi-Fi when in range of the base station and switches to LTE beyond it. Useful for farms or industrial yards with partial Wi-Fi coverage.
  • Wi-Fi mesh plus LoRa heartbeat. For multi-robot fleets, mesh Wi-Fi (802.11s or vendor-specific) handles inter-robot data while LoRa gives the operator a low-bandwidth overview of the whole fleet.

Implement the fallback logic at the application layer, not just the network layer. Your ROS 2 node should know which link is active and adjust behavior: reduce video quality on LTE, disable non-critical topics on LoRa, refuse autonomous missions if only the emergency channel is up.

How to test your link before you trust it

Datasheet numbers assume ideal conditions. Before committing a mission profile, walk the site with the robot and log link quality.

A minimal test protocol:

  1. Drive the intended route at operational speed. Log RSSI, SNR, packet loss, and latency at 1 Hz.
  2. Repeat in different weather. Heavy rain can affect longer 5 GHz links, while wet vegetation can substantially reduce Wi-Fi and LoRa range.
  3. Test with the robot in its actual pose. Antennas mounted low or shadowed by the chassis perform worse than bench tests suggest.
  4. Measure video latency end-to-end with a stopwatch and a clock on screen. Network RTT is only part of the delay.

Keep the logs. When a mission fails three months later, you will want the baseline.

FAQ

Can ROS 2 run natively over LTE?

DDS discovery does not traverse carrier NAT reliably. Tunnel ROS 2 traffic through a VPN (WireGuard, ZeroTier, or Husarnet) so both endpoints have reachable virtual addresses. Configure DDS with a static peers list or a discovery server to avoid multicast dependency.

What antenna gain should you use outdoors?

For omnidirectional Wi-Fi on a robot, 5-8 dBi is a practical balance. Higher gain flattens the radiation pattern, so if the robot tilts or the base station is on a mast, you can lose signal despite better peak numbers. For LoRa, a quarter-wave whip tuned to your band outperforms most generic antennas.

Is LoRaWAN the same as LoRa?

No. LoRa is the physical-layer modulation. LoRaWAN is a network protocol on top of it, with gateways, network servers, and duty-cycle management. For point-to-point robot telemetry you can use raw LoRa without LoRaWAN infrastructure, but regional radio regulations still apply.

How much data does an LTE-connected robot consume?

A 720p H.264 video stream at 2 Mbps consumes roughly 900 MB per hour. Add telemetry and control overhead of a few MB per hour. Plan SIM data allowances accordingly, and cap video bitrate in software to avoid surprises.

Does 5G replace Wi-Fi for robots?

Not yet, and not universally. Public 5G coverage and uplink performance vary by location and plan. Private 5G is technically strong but expensive to deploy. Wi-Fi remains the default for on-site high-bandwidth links.

What about satellite links like Starlink?

Starlink can work on stationary or mobile platforms with a clear sky view and typically provides tens to hundreds of Mbps with latency often in the tens of milliseconds, although performance varies by location and network load. Terminal power draw and physical size can make it challenging for compact rovers, but it is a valid uplink for a mobile base station serving multiple robots.

Which option is cheapest to operate?

Wi-Fi and LoRa have no recurring costs beyond hardware. LTE adds a monthly data plan, which varies by country and carrier. Over a multi-year deployment, infrastructure-based options usually cost less if you can install and maintain the base station.

To spec a rover with the right radio stack for your mission, browse configurable platforms and modules in the Fictionlab shop.


Read more