Quantum computing is a rapidly evolving field that promises to revolutionize many domains, such as cryptography, artificial intelligence, and optimization. One of the areas where quantum computing can have a significant impact is the space industry, especially in the planning of missions for Earth observation satellites (EOS).
EOS are satellites that orbit the Earth and collect data for various applications, such as weather forecasting, disaster management, environmental monitoring, and more. However, designing optimal EOS missions is a challenging task, as it involves selecting the best targets to capture from a large pool of requests, while satisfying complex constraints such as storage capacity, energy limits, weather conditions, and orbital dynamics. Moreover, the mission planning problem is dynamic and stochastic, as new requests can arrive at any time and the environment can change unpredictably.
Traditional computing methods often struggle with the complexity and uncertainty of optimizing EOS mission schedules, leading to suboptimal target selection and reduced data collection efficiency. Therefore, there is a need for new approaches that can exploit the potential of quantum computing to enhance the performance and accuracy of EOS mission planning.
A recent paper by Rainjonneau et al.1 introduces a set of quantum algorithms to solve the mission planning problem and demonstrate an advantage over the classical algorithms implemented so far. The paper uses real datasets containing thousands of tasks and multiple satellites to formulate the problem as maximizing the number of high-priority tasks completed. The paper also proposes a hybrid quantum-classical framework that combines quantum optimization and machine learning techniques to find near-optimal solutions.
The paper presents four quantum algorithms for EOS mission planning:
- Quantum annealing (QA): This is a quantum optimization technique that uses a physical device called a quantum annealer to find the lowest-energy state of a system. The paper maps the mission planning problem to a quadratic unconstrained binary optimization (QUBO) problem and uses D-Wave’s quantum annealer to solve it. The paper shows that QA can find better solutions than classical simulated annealing (SA) in terms of task completion rate and computational time.
- Quantum approximate optimization algorithm (QAOA): This is a quantum variational algorithm that uses a quantum circuit to approximate the optimal solution of a combinatorial optimization problem. The paper also maps the mission planning problem to a QUBO problem and uses QAOA to solve it. The paper shows that QAOA can outperform QA and SA in terms of task completion rate and solution quality, but requires more computational resources.
- Quantum-enhanced reinforcement learning (QRL): This is a quantum machine learning technique that uses a quantum circuit to enhance the learning capabilities of a classical reinforcement learning (RL) agent. The paper uses QRL to train an agent that can dynamically select the best targets for each satellite based on the current state of the system. The paper shows that QRL can achieve a completion percentage of 98.5% over high-priority tasks, significantly improving over the baseline greedy methods with a completion rate of 75.8%.
- Quantum kernel-based support vector machine (QKSVM): This is a quantum machine learning technique that uses a quantum circuit to compute the kernel function of a classical support vector machine (SVM) classifier. The paper uses QKSVM to classify the tasks into high-priority and low-priority categories based on their features. The paper shows that QKSVM can achieve an accuracy of 96.7% in task classification, which can help reduce the search space for the optimization algorithms.