Autonomous perimeter security: what our development vehicle video shows

The perimeters of large sites are long and empty at night, and most of the time nobody covers them.

A solar farm can have several kilometres of fence around hundreds of hectares, far from any town. A logistics park, an electrical substation or the outer perimeter of a port have the same problem: a lot of ground, few people on site and valuable material such as copper, panels or equipment. In the video in this post we show how we approach it with an autonomous ground vehicle. Here is the context.

The blind spot of long perimeters

Today a perimeter is covered in three ways, and all three have limits.

Fixed cameras cover their field of view. Covering kilometres of fence with cameras takes dozens of camera points, with civil works and communications, and some stretches are still left uncovered.

Security officers cannot be everywhere at once. Between one pass and the next there are gaps, and on a perimeter several kilometres long, shortening them means more staff and more shifts to fill.

Drones add mobility, but they are limited by flight time, weather and regulation when the job is continuous operation.

The small hours at a remote site are another gap: that is when the fewest people are available to reach the furthest stretches.

Then there is the problem that wears down whoever handles the alerts: alerts with no context. An alert that arrives without an image means checking everything, and a system like that ends up ignored. That is why every alert from the robot arrives with the image and the position, so whoever handles it can check it or dismiss it with that information.

What the system does

The video shows our development vehicle driving a route on the map in autonomous mode. It plans its path, stops for obstacles and moves on once the way is clear, with the same autonomy system as the SX-1 robot, our security robot on an industrial electric chassis.

As it drives the route, the 360° LiDAR detects people entering a detection zone and the cameras provide the image of every alert.

Beyond detection, the system marks what it has seen, a person or a vehicle, and places it on the site map. The alert reaches whoever handles it with the image and the position: the officer at your security company, your alarm receiving centre (ARC) or, on your estate, your phone. That person decides the response.

Autonomy and robot

We mount our autonomy stack, integrated and developed by us, on an industrial electric chassis.

We aim to deliver the complete robot, together with the platform and the apps each party works with, so it never turns into an integration project. For the head of security, that means a tool that fits into the operation they already run.

SeaX maintains the robots and supervises them technically. The response to each alert stays with your security company, your ARC or you.

Tested in the real world

The system has logged more than 400 km in autonomous mode on our development vehicle. We test it on a closed track and at Teruel Airport, our test site. There it faces what simulation never fully reproduces: dust, glare, uneven surfaces and conditions that change through the day. Every test session leaves data that we use in the next version of the system. The details are in how we test it.

Where we are going

An autonomous robot extends the reach of security officers to the stretches they cannot get to: perimeters several kilometres long, in the small hours.

That is what we built the SX-1 robot for. The video shows one of our tests.

How we approach it at solar farms, airports and ports and industrial sites and substations.