MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202611071177 A) filed by Noida Institute Of Engineering And Technology Niet on June 08, 2026, for Hardware-Accelerated Multi-Objective Reinforcement Learning System For Autonomous Heavy- Duty Truck Tactical Decision Making.
Inventors include Suman Rani; and Mohini Chakarverti.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: An autonomous heavy-duty truck tactical decision making system (100) comprising an onboard computing platform (200) with a field-programmable gate array fabric, a multi-objective reinforcement learning controller (300) formulating driving tasks as a multi-objective Markov decision process with vector-valued rewards for safety, time efficiency, and energy efficiency, a neural network inference accelerator (400) with latched weight registers and a fixed-point scalarization circuit producing action probabilities from per-objective action logits, a hardware safety filter circuit (500) with write-protected threshold registers implementing kinematic gap-based lane change feasibility constraints, a sensor array (600), a low-level controller interface (700), a memory protection unit (800) with a monotonic counter register, and a vehicle actuator interface (900). Upon a commit signal, the memory protection unit (800) simultaneously write-locks the weight registers, threshold registers, and policy parameter storage, establishing hardware-enforced atomic policy commitment ensuring integrity-verified operation.
Disclaimer: Curated by HT Syndication.
