<p>Robotic intelligence makes an automatic seeing environment to ravel source to target location to do automation. However, traversing path planning is a complex problem due to region coverage, obstacles, and object reasoning, which are many problems in making the decisions for optimal path planning. With billions of connected objects, managing and controlling them for large distributed networks is complex. Most traditional methods are nondependent on path planning because they take the known node point relationship of neighbor distance-based routes. The lack of behavioral analysis of feature limits leads to inadequate decision-making on the best path. To resolve this problem, we propose a novel Deep Spectral LSTM-gated convolution neural network that is used to attain a path planning decision to improve accuracy. Initially, the graph object margin is created using Relative Spectral Neighboring Graph Margin (RSNGP), and Lattice Object Convex Defense Weight (LOCDW) is used to point to an object reference. Then, the Average Moving Index Rate (AMIR) finds the mean rate of path planning to mean patterns. To the importance of the feature limits using SPIDI-Lion Optimization Algorithm (SPIDI-LOA): This finds the feature dependencies of object finding in the path to get the significance of feature relevance pattern frequency to support decision-making. Finally, LSTM-gated CNN (LSTM-GCNN) neural decisions are made by logical choices to create a linear model using an LSTM-gated unit. The proposed system achieves high performance in finding the right path to avoid congestion collision and making the best decision to choose the path compared to the other systems. This achieves the accuracy of 96% of the result on low levels of absolute error rate and high precision, as well as f1-measure accuracy compared to existing methods.</p>