Neuromorphic Perception and Navigation for Mobile Robots: A Review

TL;DR

This review assesses brain-inspired navigation for mobile robots, emphasizing asynchronous event-based perception and energy-efficient processing as realized by neuromorphic sensors and Spiking Neural Networks. It surveys localization, mapping, SLAM, planning, and control, linking them to hippocampal-entorhinal circuitry models (Place, Grid, and Head Direction Cells) and cognitive maps. The authors highlight advances in event cameras and simulations, discuss the strengths and weaknesses of current approaches, and identify key challenges such as end-to-end integration, 3D navigation, and scalable neuromorphic hardware. The work argues that combining neuromorphic sensing with brain-inspired computation could yield real-time, low-power autonomous systems, guiding future directions toward robust, 3D cognition and practical robotics applications.

Abstract

With the fast and unstoppable evolution of robotics and artificial intelligence, effective autonomous navigation in real-world scenarios has become one of the most pressing challenges in the literature. However, demanding requirements, such as real-time operation, energy and computational efficiency, robustness, and reliability, make most current solutions unsuitable for real-world challenges. Thus, researchers are forced to seek innovative approaches, such as bio-inspired solutions. Indeed, animals have the intrinsic ability to efficiently perceive, understand, and navigate their unstructured surroundings. To do so, they exploit self-motion cues, proprioception, and visual flow in a cognitive process to map their environment and locate themselves within it. Computational neuroscientists aim to answer ''how'' and ''why'' such cognitive processes occur in the brain, to design novel neuromorphic sensors and methods that imitate biological processing. This survey aims to comprehensively review the application of brain-inspired strategies to autonomous navigation, considering: neuromorphic perception and asynchronous event processing, energy-efficient and adaptive learning, or the imitation of the working principles of brain areas that play a crucial role in navigation such as the hippocampus or the entorhinal cortex.

Figures (6)

• Figure 1: Structural and hierarchical taxonomy of the survey.
• Figure 2: The underlying brain navigation cells process (adapted from Grieves_2017, samerican2016, and frontiers2021).
• Figure 3: a) Event camera pixel schematic, and b) principle of operation (from Lichtsteiner_2008). c) Time-space accumulation of events triggered by an event camera for a spinning object.
• Figure 4: Autonomous navigation flow diagram. Top: For each subsystem, the left arrows are inputs and the right ones are outputs. Bottom: examples of subsystem outputs: a) Optical flow Gehrig_2021, b) Ego-motion Vidal_2017, c) Depth ranccon2021stereospike, d) Mapping hornung13auro, e) SLAM Milford2015TowardsVS, f) Path planning Novo_2022, g) IMO detection Parameshwara_2020.
• Figure 5: Qualitative results comparing depth prediction (extracted from Hidalgo_2020). From left to right: frames, event stream, event-based predicted depth, and ground-truth depth.