Synergistic fibroblast optimization based improved reinforcement learning for intelligent assistive device

Without any human assistance, safe mobility and location identification are the two primary challenges faced by visually impaired people in the surrounding area. The affordable, reliable, robust and intelligence based assistive device is needful to blind persons for object detection and obstacle avoidance during movement. The objective of this paper is to propose a novel classical Q-learning (CQL) algorithm based intellectual walking stick prototype, using Synergistic Fibroblast Optimization (SFO) for visually impaired people. SFO algorithm is applied to choose the optimal Q-values stored in Q-table that significantly improve the learning ability of CQL for solving object detection, obstacle avoidance and location identification in complex and dynamic environments. The proposed model is able to provide effective decisions to visually challenged people as voice messages, to take safety navigational directions, based on the sensor information obtained by interacting with the environment. The prototype is exposed to learn the obstacle avoidance experience with a large volume of real time data instances collected in various real world situations, and optimized learning scheme is constructed using derived control policy. The proposed conceptual framework demonstrates an excellent generalization performance and accurate results which mean that the prototype has the capability to fulfill object detection, location identification and obstacle avoidance tasks efficiently in uncertainty environments.