The field of embodied intelligence is rapidly rising as a new focus of global attention. The just past World Robot Conference was unprecedented, with elites from all walks of life competing to show its cutting-edge technologies and innovative products, among which the humanoid robot is especially striking in its highly anthropomorphic form. However, the vast world of physical intelligence is far beyond that, and it is not the exclusive label of humanoid robots.
Physical intelligence, as the name suggests, "embodied + intelligence", by giving AI "body", can interact with reality, let AI from the software algorithm only in the digital world to the real physical world, and can present the intelligence level simulating or even surpassing human beings in the physical world. A disruptive technology that deeply integrates AI into physical entities (either humanoid robots, other robots, or other "entities"), allowing them to independently perceive, learn, and dynamically interact with their environment. But no matter what the form, the most important thing is to solve the key problem of how to move from concept to landing.
Driven by the wave of embodied intelligent technology, the field of service robots is experiencing unprecedented changes and leaps. The breakthrough of this technology not only endows service robots with more powerful perception, learning and interaction capabilities, but also greatly expands their application scenarios and potential.
1. Interaction between the body and the environment
Physical intelligence attaches great importance to the interaction between the body and the environment. Traditional artificial intelligence usually makes operations and decision-making in the virtual environment, but embodied intelligence places the agent in the real physical environment, and senses the changes of the environment with the help of various sensors, and interacts with it. It is this kind of interactivity that allows agents to more truly understand and adapt to the environment, and then make more accurate and effective decisions.
The seagull series distribution robots developed by Shuangqing Technology use cameras, lidar and other sensors to perceive the surrounding environment, identify the shape, position and other information of objects, and then carry out navigation and operation according to these information. At the same time, it can also interact with the environment through actuators like wheels, odometer and other physical level, and successfully complete various tasks. The interaction between the body and the environment enables the embodied intelligent robot to independently complete all kinds of tasks in the complex environment, which brings great convenience to the complex scenes in the work and people's production and life.
2. Coupling of perception and action
Another key feature of embodied intelligence is the coupling nature of perception and action. In the embodied intelligence system, perception and action are closely linked. The agent guides the action with the perception of the environment, and at the same time further senses the environment through the action. This coupling not only enables the agent to complete tasks more efficiently, but also improves the adaptability and flexibility of the agent. The Xiaoqing welcome robot, launched by Shuangqing Technology, will use the camera to sense the surrounding environment, identify the road, obstacles and other information, and plan the best path according to the information. At the same time, it will also execute the planned path through other actuators, continuously perceive the change in the environment in the process of action, and adjust its own action strategy. This coupling of perception and action enables the embodied intelligent robot to quickly adapt to changes in a complex environment and complete various reception tasks excellently.
3. Self-directed learning and adaptability
In addition, physical intelligence has the outstanding characteristics of autonomous learning and adaptability. The embodied agents can continue to learn and accumulate experience during the interaction process with the environment, so as to improve their own intelligence level. At the same time, it is also able to adjust its behavioral strategies to the environment to meet different task requirements. When performing a task, the Futula disinfection robot can learn how to complete the task by constantly trying and correcting error, so as to improve its own task completion efficiency. In addition, it can also adjust its behavioral strategies according to the changes in the environment, such as independently choosing to bypass obstacles when encountering obstacles. This autonomous learning and adaptability enables embodied agents to maintain efficient working conditions in a continuously changing environment and provide more reliable services for our production and life.
4. Situational and embodied
The last characteristic of embodied intelligence is situational and embodied nature. The behavior and decision-making of the embodied agent are based on the situation and physical state that it is in, and it will make different responses according to different situations and physical states. This situational and embodied nature makes the embodied agent to simulate human behavior and decision-making more realistically, and provide more natural and convenient services for human production and life. When interacting with people, the double-engine welcoming robot will give different responses according to people's facial expression, language, action and other situational information, such as smiling, nodding, and answering questions. At the same time, it will adjust its behavioral strategies according to its body state, such as charging or reducing power consumption when the battery is low.
No matter what form is adopted, the core challenge of embodied intelligence lies in how to realize the effective implementation from theoretical concept to practical application. As an emerging intelligent technology, embodied intelligence has the characteristics of interaction between the body and the environment, coupling of perception and action, autonomous learning and adaptability, situational and embodied nature. Its application is greatly accelerating the commercialization process of service robots. Through continuous optimization algorithm, improve hardware performance and deepen with the integration of all walks of life, double with robot is constantly explore the boundary of technology and application scenarios, optimize the user experience, efforts to make the robot into the medical, industrial, education, entertainment, pension, and many other fields, eventually bring human society more intelligent, convenient and efficient life experience.