Enhanced path planning for automated nanites drug delivery based on reinforcement learning and polymorphic improved ant colony optimization

摘要

Nanorobots have the potential to greatly accelerate the evolution of modern medical approaches and practices. Moreover, using artificial intelligence in a medical procedure is a rapidly growing part of today's research. Medical surgery equipment's are undergoing evolutionary advancement and microscopy procedures turn out to be an indispensable tool for nano-sample imaging, injection of medicine or surgery. In nanoscale, accurate path planning for nanites in order to reach their destination is still a challenge especially when most of the surgeries involving nanites are still being proceeded by a human operator through an interface. This article presents an algorithm capable of parallel processing of path planning with Q-learning and ACO order to reach a considerable improvement in nano-medicine delivery and optimization of path length and increasing accuracy of the results. We used autonomous path planning for post-nanite injection in vessels. We reached environmental perception and the ability to navigate quicker and more accurately in exchange with for more processing power and memory usage which will be considered an efficient trade-off (ANDD framework). The main objective of the experiments is to evaluate the performance of the proposed adaptive agent's method after it efficiently planed the path autonomously and optimized the length of the nanite swarm traveling distance. Simulation outcomes reveal that the introduced method can accomplish various objectives continuously, such as recalculation of an optimal path in case of a sudden change in patient tumor location, time efficiency in decision-making through the operation and decrease in error ratio.