An autonomous security robot developed in China provides a useful glimpse at how quickly robotics, artificial intelligence and machine vision are converging. The accompanying video shows a quadruped robot detecting a person, maintaining visual track as he moves around a building and keeping its sensor system directed towards him.
The interesting part is not the robot dog – It is the tracking.
This is no longer a remote-controlled machine where an operator manually steers a camera towards a target. The platform appears able to identify a person and maintain that track as the individual changes position.
That capability is already being marketed commercially.
From security patrol to target lock
The system is associated with Chinese technology company Nexwise and its security brand ZanX.
At the Defence Services Asia exhibition in Kuala Lumpur in April 2026, Nexwise displayed two versions of its quadrupedal unmanned ground vehicle. One was configured for autonomous patrol, the other carried a net-capture system.
According to Janes, Nexwise says the platform can autonomously detect a possible threat, alert a command centre and then lock onto the identified target. An operator can subsequently issue commands remotely.
That distinction matters.
The evidence does not show a machine independently deciding to use force against somebody.
It does show several of the technical steps immediately before that decision becoming automated:
- autonomous navigation
- obstacle avoidance
- target detection
- target recognition
- persistent tracking
- target lock
- remote command and control
- automated patrol
Nexwise also provides an API allowing different payloads to be integrated with the platform. The company told Janes that the modular design can accommodate payloads including weapon systems.
That makes this considerably more important than another robotics demonstration.
This is being sold as a security product
ZanX calls the commercial system its “Robotic Patrol Wolf”.
The company markets it for law enforcement, public security, facilities and other security roles. Its published specification describes autonomous indoor and outdoor navigation, AI-based target and behaviour recognition, automatic obstacle avoidance, remote communications and autonomous charging.
The current commercial configuration uses a non-lethal net launcher rather than a firearm. ZanX describes the system as capable of continuous autonomous patrol and 24-hour security operations.
The published hardware figures are also notable.
The Robotic Patrol Wolf weighs 59 kg, can travel at up to 4 m/s, carries up to 40 kg and has a stated battery range of 16 km. The body is rated IP67.
Those numbers put it well beyond a laboratory toy.
It is a mobile sensor platform capable of moving through spaces designed for humans, identifying activity and maintaining contact with a person without requiring an operator to continuously steer it.
The important threshold has already been crossed
Discussion about autonomous systems can become fixated on the final act of firing a weapon.
That risks missing what is happening immediately before it.
Finding a person is becoming automated.
Following them is becoming automated.
Maintaining sensor contact is becoming automated.
Classifying behaviour is becoming automated.
Navigation towards or around that person is becoming automated.
The remaining human role can become increasingly narrow.
An operator no longer needs to search manually through multiple camera feeds, identify a person, manoeuvre a platform and continually point a sensor at the target. The machine can perform much of that workload and present the operator with a tracked individual awaiting a decision.
This changes the tempo of an encounter.
It also allows one person to supervise several autonomous systems rather than directly control a single machine.
Nexwise itself describes a much broader AI business encompassing smart security, AI infrastructure and intelligent applications. The company says it has more than 1,000 staff and has participated in more than 1,000 smart-city projects. ZanX sits within that commercial technology operation rather than representing a one-off military robotics programme.
That is perhaps the most important part of the story.
The underlying technology is becoming productised.
What does this mean for military and police training?
Autonomous ground systems need to become part of the training environment.
Personnel may increasingly operate in spaces containing small robotic systems conducting surveillance, reconnaissance, perimeter security, target identification or remote intervention.
That creates a different training problem from dealing with a conventional human threat.
The machine may have detected somebody before they become aware of it. It may be transmitting their position elsewhere. Losing visual contact does not necessarily mean losing the track. Another sensor or platform may maintain it.
Training must therefore expose personnel to autonomous systems as active participants within a scenario rather than static pieces of equipment.
That means replicating autonomous patrol routes, detection behaviour, sensor coverage, classification, tracking, communications and human command relationships.
It also means testing how those systems behave when conditions become difficult: crowds, partial obstruction, poor visibility, multiple moving people, rapidly changing environments and conflicting information.
These are as much training problems as engineering problems.
Autonomy is becoming part of the operating environment
The video is striking because very little appears dramatic.
A man walks around a room > A robot watches him > The man changes direction > The robot keeps watching.
Yet behind that simple demonstration sits an autonomous chain of navigation, machine perception, classification and tracking that only a few years ago required considerably more specialist equipment.
The weapon question will attract the headlines.
The bigger development is already visible.
Machines capable of independently searching an area, identifying people and maintaining a target track are moving into commercially available security systems.
Military and law enforcement training now needs to catch up.
