Optimization of Sensor-Placement on Vehicles using Quantum-Classical Hybrid Methods

• Published in: 2022 IEEE International Conference on Quantum Computing and Engineering (QCE) pp. 820 - 823
• Main Authors: Pramanik, Sayantan, Vaidya, Vishnu, Malviya, Gajendra, Sinha, Sudhir, Salsingikar, Shripad, Girish Chandra, M, Sridhar, C V, Mathais, Godfrey, Navelkar, Vidyut
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2022
• Subjects: ansatz; Approximation algorithms; Costs; Distance measurement; Evolutionary computation; maximum set-coverage; Quantum annealing; Qubit; sensor-placement; Variational quantum algorithms; Volume measurement
• DOI: 10.1109/QCE53715.2022.00131

The paper presents a quantum method to optimize the placement of sensors on the surface of a vehicle. The problem, as posted in the BMW Quantum Computing Challenge 2021, is to arrive at the optimal positions and configurations (type and orientation) of the sensors on the vehicle surface that maximizes coverage of the Region of Interest (RoI), while minimizing the total cost of the selected sensors. The dataset contains approximately 100,000 points in the RoI, with defined measures of criticality (ranging between 0 to 1), distributed over a volume of approximately 40,000 cubic metres around the vehicle. The types of sensors, their coverage parameters and costs are inputs to the problem.The optimization problem is decomposed and solved individually for each side of the vehicle, and the results for the entire vehicle are collated and presented in the paper. The approach considers a quadratic approximation to the maximum set-coverage formulation with the advantage of obtaining the optimal number of sensors as an output from the model itself. The resultant quadratic program is solved using quantum methods, as well as a standard, state-of-the-art Integer Quadratic Program (IQP) solver, respectively. The quadratic approximation consumes qubits that are linear in the number of total configurations and significantly reduces the number of decision variables required in an Ising formulation of the problem.The quantum and classical methods are found to perform on par with each other. A detailed comparison of the results is presented in the paper. Given the combinatorial nature of the optimization problem, the classical approach may become intractable while solving the problem in an integrated manner (instead of solving individually for each side). Under such scenarios, there is a possibility to expect benefits by adopting the proposed strategies with the availability of large-sized quantum computers in the near future.