Intelligent escape and rescue are growing research areas to handle emergency situations, such as fires, earthquakes, hurricanes, and military conflicts. Biologically inspired approaches draw inspiration from advantageous biological strategies, mechanisms, and structures for the development of intelligent robotic systems that can autonomously escape threats and rescue targets. Biologically inspired approaches provide promising solutions with improved efficiency of system performance, flexibility in dynamic environments, and robustness to various uncertainties.

This thesis focuses on developing novel biologically inspired approaches to escape and rescue of multi-robot systems in dynamic and complex environments. Firstly, a novel evasion strategy is designed for multiple evaders against a faster pursuer in dynamic environments. A neurodynamics-based approach is proposed to approximate the pursuit-evasion game, instead of differential games, which can provide real-time responses to sudden changes in complex environments. Secondly, a novel fish-inspired collective escape approach is developed for multi-robot systems to leave away from threats with limited sensing ability. The proposed neurodynamics-based self-adaptive mechanism enables multi-robot systems with the self-adaptive ability in responding to environmental changes. Finally, a novel feature learning-based bio-inspired neural network (FLBBINN) is proposed to quickly generate a heuristic rescue path in complex and dynamic environments to improve the effectiveness and efficiency of multi-robot systems. Extensive real-robot experiments are conducted to verify the performance of the proposed approaches in real-world environments.