How to ensure the physical safety of humanoid robots during human-robot interaction?

We can discuss this topic from several angles, much like setting up a comprehensive security system for a home, which requires a multi-faceted approach encompassing hardware, software, and external regulations.

How to Ensure Physical Safety of Humanoid Robots in Human-Robot Interaction?

Imagine a new member joining your family – a humanoid robot. It can help with chores and chat with you, but it's ultimately a large machine made of metal and motors. How can we ensure that while it provides convenience, it doesn't accidentally harm us due to its 'clumsiness'?

This is essentially a systems engineering challenge, which can be broken down into the following key areas:

1. Making its 'Body' Soft (Hardware Design)

This is the most intuitive line of defense. If the robot itself is 'soft,' then even in the event of a collision, the harm will be significantly reduced.

Lightweight Materials: Just as cars strive for lightweight designs, robots also use lighter yet sufficiently robust materials. With less mass, the impact force during a collision naturally decreases.
Rounded Design: You'll notice that modern humanoid robot designs tend to be rounded with no sharp edges. This is similar to how we put corner protectors on furniture, avoiding direct pokes and scratches.
'Compliant Joints': This point is crucial. Traditional industrial robot joints are 'rigid,' stopping and moving abruptly with great force. Modern humanoid robots, however, incorporate cushioning structures similar to 'springs' or 'airbags' within their joints (professionally known as 'Series Elastic Actuators' or 'Compliant Control'). When its arm touches you, this cushioning structure absorbs most of the impact, making it feel like a gentle push from a pillow, rather than being poked by an iron rod.

2. Equipping it with Sensitive 'Eyes' and 'Skin' (Sensor Technology)

A 'soft' body alone isn't enough; robots also need to constantly perceive their surroundings and the forces they interact with, to prevent collisions at the source.

3D Vision ('Eyes'): Through devices like cameras and LiDAR, the robot constructs a three-dimensional virtual world in its 'mind.' It can clearly 'see' where you are and where furniture is, allowing it to plan a safe path that avoids any contact when moving its arms or body.
Force/Torque Sensors ('Skin'/'Touch'): Highly sensitive force sensors are installed in every joint, and even on the outer shell, of the robot. This means the robot has a 'sense of touch.' If its arm, during movement, detects any unexpected slight contact (e.g., your hand suddenly blocking its path), it will instantly recognize this 'anomaly' and immediately cease movement.
Proximity Sensors: Some key parts of the robot (e.g., palms, chest) will have close-range sensors. When an object (like your face) approaches a dangerous distance, it will forcibly decelerate or stop. This serves as the final line of defense.

3. Teaching it to 'Look Before It Leaps' (Software and Algorithms)

With a 'body' and 'senses,' a smart 'brain' is also needed to process information and make correct decisions.

Safe Path Planning: Before executing any action, the robot's 'brain' first simulates it in its virtual world. It calculates a path that not only accomplishes the task but also maximizes distance from humans.
Human Intent Prediction: More advanced robots can even 'predict' your behavior. By observing your posture, gaze, and movements, they can anticipate where you might go next or what you might pick up, thereby actively avoiding you or collaborating with you in a safer manner.
Adhering to the 'Three Laws of Robotics': While this is a concept from science fiction, engineers in reality are turning it into code. The robot's core programming includes unbreakable 'red lines,' such as: under no circumstances should the force exerted on a human exceed a certain safety threshold; when a potential collision with a human is detected, the highest priority command is to 'stop' or 'retreat,' rather than to 'complete the task.'

4. Establishing Strict 'House Rules' (Standards and Regulations)

Finally, just as all products must adhere to national standards, robots also require a unified 'safety manual.'

Industry Safety Standards: Both international and domestic bodies have specific safety standards for collaborative robots and humanoid robots (e.g., ISO/TS 15066). These standards detail restrictions on speed, force, design, and other aspects when robots work alongside humans. Manufacturers must comply with these standards for their products to be released to the market.
Emergency Stop Button: This is the simplest and most reliable safeguard. A very conspicuous red 'emergency stop' button will be located on the robot itself or in its working environment. No matter what the robot is doing, pressing this button will instantly cut its power and halt all movements.

In Summary

Ensuring the physical safety of humanoid robots isn't reliant on a single 'black technology,' but rather on a multi-layered safety system composed of 'compliant hardware + sensitive sensing + intelligent algorithms + strict standards.' Each layer acts like a sieve, filtering out potential dangers, ultimately ensuring we can coexist safely and comfortably with these intelligent 'big guys.'