MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202611072638 A) filed by Noida Institute Of Engineering And Technology Niet on June 11, 2026, for Edge-Centric Graph Neural Network Accelerator System For Energy-Aware Electric Fleet Route Optimization.
Inventors include Rohit Chaudhary; and Dr. Vivek Kumar.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: An edge-centric graph neural network accelerator system (100) for energy-aware electric fleet route optimization comprises an edge-attention encoder circuit (110) processing edge feature vectors of a road network graph through stacked edge-attention layers to produce node embeddings capturing travel costs and energy consumption, an attention-based route decoder circuit (120) constructing vehicle routes using composite attention scores incorporating semantic compatibility and energy penalty terms, a hardware-enforced feasibility masking controller (130) implemented as an interconnect firewall circuit with write-locked constraint policy registers interposed between score computation and softmax computation to block constraint-violating action selection, a battery dynamics tracking unit (140) maintaining per-vehicle state-of-charge registers updated after each edge traversal with charge increments computed from a charging profile lookup table in a write- protected memory region (150), and a fleet orchestration state machine (160) coordinating multi-vehicle dispatch through hardware-latched fleet status registers.
Disclaimer: Curated by HT Syndication.
