Introduction
The use of Artificial Intelligence (AI) in the space domain has only just begun, and the industry is expecting a plethora of innovations with such advancements in the coming decades. While the role of AI hasn’t been fully explored, Google’s CEO Sundar Pichai once stated, “Artificial Intelligence will have a more profound impact on humanity than fire, electricity, & the Internet.” This statement signals that AI will bring about a whole new revolution, profoundly impacting humanity’s future.
To understand what AI is, its applications in the space industry, and what the future looks like, we will dissect this topic step by step in the following sections of the article.
What is AI?
AI is a branch of computer science that deals with making computers think in an intellectual manner similar to the human brain, deriving results based on sample datasets, research, and observing patterns. Simultaneously, Machine Learning (ML) is a branch of Artificial Intelligence that focuses on using data and algorithms to enable AI to learn and imitate the way humans learn a task and gradually improve accuracy.
Currently, AI is being used extensively in different industries such as IT, Robotics, Healthcare, Finance, and Entertainment, and the Space industry is one of them.
Specific to the space industry, AI can be used in autonomous navigation and operations, data analysis and management, mission planning and optimization, enhancing communication systems, human-AI collaboration, and much more.
NASA’s footprint in AI
In 2017, Google and NASA built an AI algorithm to sift through data from the Kepler mission. The program found two exoplanets that humans had missed. The newly discovered Kepler-90i, a sizzling hot rocky planet that orbits its star once every 14.4 days, was found using ML from Google. In this case, computers learned to identify planets by finding instances in Kepler data where the telescope recorded signals from planets beyond our solar system, known as exoplanets. They claim that more than humans, AI is great at identifying false positives.
Furthermore, NASA developed a program to create 3D models for asteroids, a method that helps detect potentially hazardous asteroids. This work can be completed in four days, a task that could take months.
Understanding the AI and ML dynamics
AI and ML need historical data to make predictions, find patterns, and perform analytical work. This historical data can be used in the requirement analysis and system design phases of space software development, where user feedback on previous designs, previous software performances, and historic mission data can be utilized. It is also important to consider the historical data of failed missions, which can help establish critical measures to follow while gathering requirements and creating fault-tolerant space software designs. Negligence can cause significant losses, so precision in historical data and its analysis is crucial.
In Space software development, anomaly detection is one of the key aspects of mission safety. AI and ML can be used to predict anomalies based on available data. We can use telemetry data, which consists of continuous streams of data from spacecraft sensors (including temperature, pressure, and power usage), to monitor failure predictions and health. System logs can be used to analyze software operation errors and system events. By performing predictive analysis on them, we can avoid future unwanted and predictable errors. Command and control data, which is information about commands sent to the spacecraft and their execution results, can be very helpful in monitoring and maintaining spacecraft health.
Unforeseen risks
Humans are prone to error while completing a task; on the other hand, machines can complete tasks with greater precision. However, the irony is that the software, which is the core of the machine, is also developed by humans.
Let us consider an AI blunder from the aviation industry as an example. In February 2024, Air Canada had to pay damages for misinformation provided by a chatbot. One of the consumers, Jake Moffatt, consulted Air Canada’s virtual assistant about bereavement fares after his grandmother passed away in November 2023. The chatbot informed him he could buy a regular-priced ticket from Vancouver to Toronto and apply for a bereavement discount within 90 days of purchase. Following this advice, Moffatt purchased a one-way CA$794.98 ticket to Toronto and a CA$845.38 return flight to Vancouver. However, when Moffatt submitted his refund claim, the airline turned him down, stating that bereavement fares couldn’t be claimed after tickets were purchased. Moffatt took Air Canada to a tribunal in Canada, claiming the airline was negligent and misrepresented information via its virtual assistant, and he eventually received a refund on his tickets.
Referring to the above incident, what if an astronaut asks an AI chatbot, “Can you check the status of the life support system?” and the system responds incorrectly based on previous historical data due to a technical data syncing issue? In space missions, precision is crucial. With such high demands for accuracy, can AI truly be trusted in space missions? This is a question the industry needs to address.
On the other hand, in the healthcare industry, since the COVID-19 pandemic began in 2020, numerous organizations have sought to apply ML algorithms to help hospitals diagnose or triage patients faster. However, according to the UK’s Turing Institute, a national center for data science and AI, the predictive tools made little to no difference. MIT Technology Review has chronicled several failures, most of which stemmed from errors in the way the tools were trained or tested.
The use of mislabeled data or data from unknown sources was a common issue. Derek Driggs, an ML researcher at the University of Cambridge, along with his colleagues, published a paper in Nature Machine Intelligence exploring the use of deep learning models for diagnosing the virus. The paper concluded that the technique wasn’t fit for clinical use. For example, Driggs’ group found their own model was flawed because it was trained on a data set that included scans of patients lying down while being scanned and patients standing up. Patients lying down were much more likely to be seriously ill, so the algorithm learned to identify COVID risk based on the position of the person in the scan. Similarly, an algorithm trained with a data set that included scans of the chests of healthy children learned to identify children as not high-risk patients.
Space environments are harsh and unpredictable. AI systems need to be extremely reliable and capable of handling unexpected situations without human intervention. In error handling, AI algorithms must be designed to manage errors accurately, as the consequences of mistakes in space can be severe.
Conclusion
As new technological advancements emerge, AI and ML will continue to grow in space software development. Furthermore, other space software advancements, such as automatic data visualization dashboards, could be utilized to present telemetry data more meaningfully using visualization techniques like heatmaps, trend graphs, and anomaly plots, helping operators quickly understand system status and avoid unwanted events.
Future enhancements might include improved human-robot interactions, space traffic management, self-healing systems, AI-driven mission planning, and more. While AI systems have proven well-suited for repetitive tasks in harsh and hazardous environments, there is a potential mismatch between AI’s current ability in perception and the intelligent decision-making required by humans for complex tasks in space. The harsh and complex space environment is the biggest challenge when it comes to training AI models to behave and make accurate predictions; even a minor logical error could result in irreversible damage. Therefore, as the use of AI rapidly grows in space missions, mapping risks remains a crucial task for the industry.
Author
Aishwarya Kasulkar is a dedicated data engineer with a passion for harnessing the power of data to drive innovative solutions. With a strong background in cloud technologies and data engineering, she excels at transforming complex datasets into actionable insights. Beyond her professional expertise, Aishwarya is a space enthusiast who is fascinated by the mysteries of the universe and the role of technology in space exploration. She enjoys exploring the intersection of data science and space technology, constantly seeking new ways to contribute to advancements in the field. She can be reached at: kaishwaryasubhash@gmail.com.
