14 - NeuroRover

Inspiration

Our project, NeuroRover, was inspired by the growing need for autonomous systems that can patrol and monitor border areas, especially in regions where terrain is difficult or constant human surveillance is not feasible. We envisioned a robotic scout capable of navigating these areas intelligently, using bio-inspired neuromorphic systems to simulate human-like decision-making. The goal is to enhance situational awareness and security while reducing the risk to human personnel.

What it does

NeuroRover is a simulation of a robotic scout that uses a neuromorphic (brain-inspired) neural network to autonomously patrol a defined area—such as a border zone—while avoiding obstacles and adapting its path in real time. The robot mimics the behavior of a border patrol unit, using spiking neural network responses to assess its environment through simulated sensory input. It dynamically adjusts its route, responds to potential threats or barriers, and ensures comprehensive area coverage with minimal human input.

How we built it

We used Brian2, a Python library for spiking neural network simulations, to model the patrol robot’s "brain." The system receives input from simulated proximity and environmental sensors and processes that information through voltage-based neuron models. These neuron-like units generate movement commands to guide the robot along a patrol path while actively avoiding hazards. Although the model remains a simulation due to hardware limitations, it effectively represents a foundation for real-world scout robotics in security or military contexts.

Challenges we ran into

Adapting the original navigation system to a continuous patrol route—rather than a point-to-point target—presented new challenges. We had to rework the neural logic to support patrol behaviors such as area scanning and route looping. Translating this behavior into realistic spiking neuron dynamics also required careful tuning of neuron thresholds, timing, and interaction between simulated sensors and the neural controller.

Accomplishments that we're proud of

We're proud of how we adapted the neuromorphic system for a security-focused application, expanding our understanding of both patrol logic and neural modeling. Collaborating as a team to simulate a functional patrol robot with biologically-inspired behavior was a key milestone for us.

What we learned

We learned how neuromorphic networks can be adapted to different behavioral models beyond goal-seeking navigation. We deepened our understanding of the Brian2 library, sensor-to-neuron signal processing, and how to simulate adaptive behaviors like scanning and obstacle evasion within a patrol scenario.

What's next for NeuroRover

The next step is to integrate our simulation into a physical robotic platform for real-world testing in controlled environments. This will allow us to assess how well the patrol logic holds up under actual terrain conditions and how it might contribute to real-time border surveillance or autonomous monitoring systems.

Built With

• brian2
• python