Innovative robot navigation inspired by brain function boosts efficiency and accuracy
A QUT research team has taken inspiration from the brains of insects and animals for more energy-efficient robotic navigation.
Led by postdoctoral research fellow Somayeh Hussaini, alongside Professor Michael Milford and Dr Tobias Fischer of the QUT Centre for Robotics, the research, which was published in the prestigious IEEE Transactions on Robotics journal and supported by chip manufacturer Intel, proposes a novel place recognition algorithm using Spiking Neural Networks (SNNs).
“SNNs are artificial neural networks that mimic how biological brains process information using brief, discrete signals, much like how neurons in animal brains communicate,” Miss Hussaini said.
“These networks are particularly well-suited for neuromorphic hardware—specialised computer hardware that mimics biological neural systems—enabling faster processing and significantly reduced energy consumption.”
While robotics has witnessed rapid progress in recent years, modern robots still struggle to navigate and operate in complex, unknown environments. They also often rely on AI-derived navigation systems whose training regimes have significant computational and energy requirements.
“Animals are remarkably adept at navigating large, dynamic environments with amazing efficiency and robustness,” Dr Fischer said.
“This work is a step towards the goal of biologically inspired navigation systems that could one day compete with or even surpass today’s more conventional approaches.”
The system developed by the QUT team uses small neural network modules to recognise specific places from images. These modules were combined into an ensemble, a group of multiple spiking networks, to create a scalable navigation system capable of learning to navigate in large environments.
“Using sequences of images instead of single images enabled an improvement of 41 per cent in place recognition accuracy, allowing the system to adapt to appearance changes over time and across different seasons and weather conditions,” Professor Milford said.
The system was successfully demonstrated on a resource-constrained robot, providing a proof of concept that the approach is practical in real-world scenarios where energy efficiency is critical.
“This work can help pave the way for more efficient and reliable navigation systems for autonomous robots in energy-constrained environments. Particularly exciting opportunities include domains like space exploration and disaster recovery, where optimising energy efficiency and reducing response times are critical,” Miss Hussaini said.
Read the full paper, Applications of Spiking Neural Networks in Visual Place Recognition, online.
Main image (left to right): Dr Tobias Fischer, Somayeh Hussaini, Professor Michael Milford
Media contact:
Lauren Baxter
QUT Media
media@qut.edu.au
07 3138 2361 / 0407 585 901 (After Hours)
