Quantum-Enhanced Hybrid Reinforcement Learning Framework for Dynamic Path Planning in Autonomous Systems

  • 2025-04-29 12:36:08
  • Sahil Tomar, Shamshe Alam, Sandeep Kumar, Amit Mathur
  • 0

Abstract

In this paper, a novel quantum classical hybrid framework is proposed thatsynergizes quantum with Classical Reinforcement Learning. By leveraging theinherent parallelism of quantum computing, the proposed approach generatesrobust Q tables and specialized turn cost estimations, which are thenintegrated with a classical Reinforcement Learning pipeline. The ClassicalQuantum fusion results in rapid convergence of training, reducing the trainingtime significantly and improved adaptability in scenarios featuring static,dynamic, and moving obstacles. Simulator based evaluations demonstratesignificant enhancements in path efficiency, trajectory smoothness, and missionsuccess rates, underscoring the potential of framework for real time,autonomous navigation in complex and unpredictable environments. Furthermore,the proposed framework was tested beyond simulations on practical scenarios,including real world map data such as the IIT Delhi campus, reinforcing itspotential for real time, autonomous navigation in complex and unpredictableenvironments.

 

Quick Read (beta)

loading the full paper ...