Fences, badge readers and CCTV walls were built for a world in which physical and digital threats lived apart. In a column published on The Robot Report on October 4, 2026, security and tech writer Charlie Burgess argues that this world is gone, and that robotics paired with physical AI is one of the few tools that can keep up. His example is a four-legged patrol robot from Foxlink and Luminys, built on NVIDIA and SYNC ROBOTIC technology.
The money is moving. According to the column, more than $23 billion has been invested in 2026 alone in organizations working on how robots and AI function together.
Three problems old security can’t solve
Burgess sorts the risks to data centers, factories and other critical sites into three groups:
- Cyber-physical convergence: cameras, locks and sensors are now networked, so a single coordinated attack can hit both the digital and the physical layer.
- Legacy constraints: older hardware and software leave security teams reacting to incidents rather than getting ahead of them.
- Global volatility: geopolitical and socioeconomic tension is raising the threat level across the board.
To show what convergence looks like in practice, the column points to the 2025 attack on Jaguar Land Rover. Intruders got into the carmaker’s IT systems using stolen credentials and social engineering. JLR confirmed the incident on September 2, 2025, and shut its systems down to contain it. Production stayed suspended until at least October 1, 2025. Burgess puts the cost to the British economy at an estimated $2.5 billion.
A patrol dog with an NVIDIA brain
The column’s answer is the Foxlink and Luminys quadruped, which it describes as a next-generation security asset. Smart sensors let it monitor an area, while physical AI on board looks for anomalies instead of just recording footage for someone to review later. The companies first showed the robot at GTC Taipei on May 27, 2025. According to that announcement, the platform offers:
- facial recognition;
- modular sensors, including heat detection;
- SLAM-based autonomous navigation;
- LLM integration for adaptive decision-making;
- long-horizon memory through NVIDIA ReMEmbR.
Under the hood is a large share of NVIDIA’s robotics stack: Jetson AGX Orin, GR00T N1, Isaac Sim, TensorRT, Metropolis, X-Mobility and RIVA. The robot was also trained in Omniverse, where it practised virtual parkour.
On June 17, 2025, Luminys followed up with SYNC ROBOTIC’s AI Robot Security Inspection Solution. It can run on either quadruped robots or drones and offers 24-hour autonomous inspection, automatic charging, task scheduling, modular detection sensors and real-time video streaming. “This launch demonstrates NVIDIA and Foxlink’s strategic collaboration, opening a new chapter in global smart security,” said Freddy Kuo, chairman of Luminys and SYNC ROBOTIC.
Augment, don’t rip and replace
Burgess’s main point is about integration, not hardware. Robots like this should plug into existing legacy systems rather than replace them, so they fill coverage gaps that fixed cameras miss. Large language models then learn from threats and analyze patterns across the whole estate. The column says such systems are already in use at hospitals, shopping malls and education campuses around the world. Its argument is that adaptive, scalable security is now something operators can deploy, not just discuss.
