A Bio-Inspired and Solely Vision-Based Model for Autonomous Navigation

Vision, utilizing eyes or cameras as the predominant sensory input, emerges as the primary and informative perception source for both animals and robots, facilitating crucial functions such as perception, navigation, interaction, and comprehensive understanding of their surroundings. Inspired by insect-like invertebrate animals’ remarkable visual processing abilities which are achieved despite their constrained computational resources, this paper delves into the realm of bio-inspired autonomous navigation. The objective is to tackle the substantial challenges of cost efficiency, ensuring a solely vision-based algorithm guides the agent to its destination without collisions. To this end, a vision-guided navigation model is proposed by orchestrating the neural model of ant’s visual navigation and crab’s looming spatial localization. As the exploratory endeavor in constructing a bio-inspired autonomous visual navigation model, which demands significantly lower computational resources in comparison to prevailing engineering solutions. The effectiveness of this model has been validated through rigorous systematic experiments, lending empirical support to its capabilities. The proposed vision-motion closed-loop framework imparts valuable insights for the development of more efficient and precise autonomous systems, embodying the principle of drawing inspiration from nature’s wisdom.