Enceladus is one of Saturn’s satellites, and under its thick, icy surface lies an ocean of liquid water. For this reason, astronomers believe if there’s life in the Solar System outside of Earth, this will be where it is. The ocean is covered by a thick layer of ice, so the exploring robot can’t communicate with engineers while in the ocean. If something goes wrong, the robot has to find solutions by itself. Nobody knows what dangers this lurk in the depths of Enceladus, radiation, engine failure, or something else… That’s why we believe that solution is using a self-learning algorithm to make decisions in the space that can find creative solutions to problems and work like human pilots. However, this kind of algorithm has lots of issues and risks. This is where our project Polimane steps in and makes this artificial intelligence possible with video games.
Landing on Enceladus
More about The Mission
Reaching Enceladus and landing on it safely can be achieved with pre-programmed commands like other spacecraft missions. However, all of the other operations will be controlled by a group of machine learning algorithms that use reinforcement learning to adapt to new situations and environments.
Traveling on the Surface
Problems with Reinforcement Learning
Reinforcement learning can make agents learn the most optimal solutions to problems in unique environments through trial and error. However, the agent needs to make mistakes in the environment and need lots of time to adapt to solve problem. Unfortunately, in space missions robots neither have time nor the luxury of making mistakes. That’s why reinforcement learning algorithms like Q-Learning need a better and faster way to learn without making serious mistakes. Project: Polimane will allow human players’ experience in gaming and curiosity to prepare Q-Learning to solve this problem.
Purpose of This Mission
For Space Missions
Success in space missions requires the collaboration of thousands of people with different professions, years of work, and decades of experience. For example, NASA JPL’s Mars missions how this progress very well because, with each new mission, they had harder and bigger goals step by step. However, for missions to planets farther than Mars, spacecrafts need years to reach target locations. Because of this long time period, progress and gaining experience will be slower. Also, sending data takes longer, so as spacecrafts go farther, they become more alone. That’s why we believe that our ultimate goal of creating artificial intelligence that has high reliability and adaptation skills with human curiosity and experience has the potential of becoming an essential component of space missions. Using these kinds of artificial intelligence in space missions is unusual, but we think it will be essential to start a new era in humanity’s place in the universe.
For Artifical Intelligence
According to most researchers, artificial intelligence lives its golden age because every day, new research comes that shows the new amazing capability of artificial intelligence. However, researchers started to worry that artificial intelligence may reach its limit because of processing power and hardware. Also, the idea that deep learning’s core backpropagation needs to be replaced gets more popular every day. There are different solutions ideas, and our idea is that artificial intelligence needs to experience and critical thinking ability that humans earn with actually living in an environment for years. Also, it is a common fear that artificial intelligence will cause more worklessness, more ethical issues, and even the extinction of humanity.
That’s why we are working on a video game that will allow everyone to use their unique experiences and perspectives to train artificial intelligence(A. I. ) models. This will make humans very essential in training artificial intelligence, so worklessness caused by A. I. can be fixed with this job type. Creating a closer relationship between A. I. and humanity with making A. I. more understandable by everyone will be very helpful in creating a peaceful environment both for A. I. and humans. There are also many potential benefits for different areas and problems like solving mathematical paradoxes and getting closer to the P=NP problem. More about this can be found on the page about Project: Polimane.
Robot Dodo is a 4-wheel robot that can go underwater autonomously using Q-Learning supported by our video game Project: Polimane and making different measurements. It is designed as a simplified version of the underwater robot that is planned to explore the hidden ocean of the Enceladus. The mission of this robot is to explore the underwater by finding solutions to unexcepted problems like one of the motors stopping working. With artificial challenges created by us and existing natural challenges, Project: Polimane’s capability to create adaptive artificial intelligence models will be tested in real life.