Multi-Path Communication for a Military Reconnaissance Robot

The Project

The Orpheus-AC3 is a CBRN (chemical, biological, radiological, nuclear) reconnaissance robot deployed by the Czech Army. 40 units are in active service as part of the LOV-CBRN vehicle system, delivered between 2021–2023. The goal of this project was to develop the next generation — Orpheus-XR — with substantially improved optoelectronic and communication capabilities.

The project was funded by the Technology Agency of the Czech Republic (TAČR) under Programme TREND, with a consortium of four partners:

• LTR s.r.o. — lead partner, robot manufacturer
• Brno University of Technology (VUT) — optoelectronics, sensor fusion, telepresence
• T-Mobile Czech Republic — private 5G SA network, coverage mapping
• GK SERVIS — communication subsystem development and integration

GK SERVIS’s Role: Communication Subsystem

GK SERVIS was responsible for Stage 5: Development and Integration of the Communication Subsystem. The core challenge: a reconnaissance robot must maintain unbroken connectivity to its operator in environments where no single communication technology is reliable — dense urban areas, underground spaces, forests, disaster zones, and areas with intentional signal jamming.

We developed a Multi-Path Connectivity strategy that automatically aggregates and switches between available communication layers based on real-time link quality assessment.

5G integration (private + public)

In partnership with T-Mobile, we optimized the system for operation on a private 5G SA (Standalone) network in the n78 band. This provides ultra-low latency and high throughput for real-time HD video streams and VR telepresence. When private coverage is unavailable, the system seamlessly falls back to public 5G/LTE infrastructure.

We tested multiple 5G chipsets and found Qualcomm-based modules delivered approximately 2x the throughput of SIMCom alternatives under identical radio conditions — critical for streaming multi-camera feeds in real time.

MANET mesh network

For infrastructure-free operation (non-line-of-sight, deep forests, building interiors), we integrated Persistent Systems MPU5 modules using Wave Relay technology. The robot acts as a mobile relay node, dynamically extending network reach with each additional node in the field. The system supports switching between civilian C-band and military L-band frequencies.

We also tested Doodle Labs mini-OEM Dual-Band Mesh Rider Radio modules for additional mesh networking scenarios.

Satellite connectivity

For operations in complete “white spots” with zero terrestrial coverage, we tested two satellite solutions:

• Starlink (LEO) — tens to hundreds of Mbps with sub-50ms latency. Enabled 4K video streaming from the robot in remote areas. Limited by line-of-sight requirements and brief dropouts near tall obstacles (compensated with buffering + MANET fallback).
• Inmarsat Class 7 (GEO) — higher resilience to weather and broader geographic coverage. Used as a critical backup channel for telemetry, GPS coordinates, and emergency return commands when all other links fail.

Software-Defined Radio (SDR)

Unlike fixed-architecture modems, the SDR module can dynamically change frequency and modulation type in real time. This is critical when standard bands (WiFi, 5G) are jammed or intentionally blocked. The SDR also bridges communication with legacy analog systems and specialized sensors that don’t support modern IP protocols.

Operator station communication module

We designed and built a portable communication/retranslation module for the operator station, featuring its own swappable battery, integrated RTK-GNSS receiver for precision positioning corrections, and all communication interfaces needed to maintain the multi-path link with the robot.

The Robot: Orpheus-XR

The Orpheus-XR carries a sensor head with 2 RGB cameras, 2 thermal cameras, and a 3D LiDAR (Livox MID-360), all connected through a custom FPGA + Jetson Orin processing board. The communication subsystem we built handles the enormous data throughput from these sensors — streaming high-resolution video, thermal imagery, and 3D point clouds simultaneously to the operator.

The system includes adaptive transmission: when link quality degrades, it automatically adjusts resolution, frame rate, and compression (JPEG, H.264, H.265) per sensor, or temporarily disables individual sensors to free bandwidth. Latency is continuously monitored at every stage of the pipeline — from capture to compression to wireless transmission to display.

Testing Methodology

Beyond the Orpheus-XR itself, the consortium built two additional test robots (Scout-mini I & II) specifically for communication experiments. These robots run an autonomous waypoint-following system (developed in Python) with RTK-GNSS precision, allowing repeatable outdoor communication measurements across different configurations, times, and conditions.

The VUT campus area was equipped with 16 geodetically surveyed reference points and a rooftop RTK-GNSS base station, creating a permanent outdoor testbed for radio transmission experiments.

Results

• Unbroken connectivity across urban, indoor, underground, and open-field environments through automatic multi-path switching
• 4K video streaming from remote areas via Starlink with sub-50ms latency
• Dynamic mesh extension — each robot in the field acts as a relay, extending range with every additional node
• Adaptive bandwidth management — automatic sensor resolution/compression adjustment based on real-time link quality
• Anti-jamming capability through SDR frequency hopping
• Portable operator module with integrated RTK-GNSS and swappable battery for field deployment

Key Takeaway

In critical robotics applications — whether military reconnaissance, disaster response, or industrial inspection — communication is the single point of failure. A robot that loses its link to the operator is worse than useless; it’s a liability.

The multi-path architecture we developed eliminates this single point of failure. No single technology is reliable everywhere, but the right combination of technologies, with intelligent failover, can be. That’s what this project delivered.