Embodied cognitive intelligence guided Moon sample collection

The success of Chang’e-61,2 is a milestone of lunar exploration by China, being the first successful attempt in human history to collect samples from the far side of the Moon. To ensure collection efficiency and success of the mission, operators on the Earth needed to cooperate with the sample collection equipment on the Moon and finish the sample collection in less than a day. Embodied cognitive intelligence (ECI), which dynamically combines the advantage of human cognition with embodied intelligence, fitted this mission well. Experts on the Earth accumulated experience and knowledge to guide the design of the Chang’e probe and also conducted extensive simulation experiments for various challenges arising on the Moon. We used these accumulated data and experience along with knowledge to construct a cognitive map and then use the real-time data of Chang’e-6 in an embodied way by intelligently interacting with the environment. By using ECI, we successfully collected samples on the far side of the Moon in a fast, accurate, and robust way.

Moon’s exploration by China

In 2013 Chang’e-3, equipped with a robotic arm, landed on the Moon. In 2018–2019, Chang’e-4 landed on the far side of the Moon. In 2020, the Chang’e-5 probe collected 1,731 g of sample on the near side of the Moon. In 2024, Chang’e-6 was launched and collected 1,935.3 g of sample on the far side of the Moon. The Chang’e-7 and Chang’e-8 probes will be launched in the next few years. An International Lunar Research Station (ILRS) has also been planned for long-term lunar exploration.

Embodied cognitive intelligence

The long-term study of mammals indicates the usefulness of cognitive mapping3 for environment cognition, interpretation, and adaptation. The interaction between brain and environment provides rich experience and basic knowledge for cognition. Cognitive mapping can gradually adapt to the changing environment. Embodied intelligence4 tries to integrate physical interaction with artificial intelligence (AI) in real-world scenarios, which is a possible path to artificial general intelligence. It has shown potential in various applications (e.g., robotics, autonomous driving, and intelligent manufacturing). However, it also faces shortcomings such as the requirement of rich data and high computational cost. Besides, its performance is still unsatisfactory for high-cost and dangerous applications (e.g., lunar exploration and toxic chemical waste disposal). Human experience and knowledge should additionally be used.