Please note that English isn’t my native language, I had help from ChatGPT to translate those essays to English, and thanks for your reading.
Reflections on Optimizing Robotic Balance Through Hardware Design (Part 3)
In most people, when performing a twisting motion—such as rotating the torso left or right—the body’s center of gravity (COG) doesn’t remain perfectly centered between the feet. Instead, it subtly shifts left or right, depending on the direction of the movement.
Kung Fu as a Frame of Reference
To illustrate this concept, I’ve included an image of a Kung Fu performance. Martial arts provide a compelling example for understanding dynamic balance because they emphasize deliberate weight shifting, controlled momentum, and refined postural adjustments.
When a performer twists their torso to the right, the weight distribution between their feet might shift, for instance, 30% on the left foot and 70% on the right. Conversely, when twisting to the left, the distribution flips. Such subtle yet deliberate shifts allow the body to maintain balance and prepare for the next movement.
However, these weight shifts are often imperceptible when analyzing a static photograph or observing external postures. Weight distribution isn’t limited to left and right feet; there are further nuances:
- Weight can shift toward the heel, across the midfoot, or onto the toes, depending on the intended movement.
- The timing, speed, and force with which the COG shifts are critical.
To borrow imagery from martial arts or wuxia novels: even if two martial arts masters stand perfectly still, one may lose to the other due to a superior control of “hidden force” or weight readiness—invisible adjustments that enable explosive movement.
The Concept of “Momentum Readiness”
All humans who walk intuitively achieve a phenomenon I call “momentum readiness.” Before taking a step forward, your body naturally shifts its weight in preparation. The key point here is:
- The foot may not have moved yet, but the COG has already begun to reposition.
- This repositioning allows for an effortless transition into the next step.
This concept also explains why household pets—like cats—so often trip people. Imagine this scenario:
- Just as you are about to step forward, your body has already shifted its weight to prepare for the movement.
- Suddenly, a cat brushes against your leg, forcing you to halt mid-motion or readjust your balance.
- Even if your foot has only lifted two centimeters off the ground, this sudden disruption activates a flurry of instinctive balancing reflexes—flailing arms, shifting weight, and awkward steps—to avoid falling.
Implications for Robotic Design
To improve the balance and movement efficiency of robots, engineers and scientists must focus on the pre-movement adjustments that humans instinctively perform. These adjustments involve:
- Dynamic Center of Gravity Shifts
- Robots should be able to anticipate and reposition their COG before initiating a movement.
- This would reduce “starting resistance” or the force needed to overcome inertia (akin to static friction in physics).
- Weight Redistribution Mechanisms
- Robots could benefit from systems that allow real-time weight shifting (e.g., adjustable actuators or internal weight-balancing systems).
- These systems could emulate how humans reposition weight across the feet, whether toward the heels, midfoot, or toes.
- Pre-Movement State Awareness
- Sensors and AI algorithms should detect the robot’s current balance and anticipate weight adjustments necessary for a smooth transition into motion.
- Just as humans unconsciously prepare their weight before stepping, robots could simulate this “momentum readiness” for more fluid movement.
Conclusion: Moving Beyond Static Designs
Robotic motion cannot be fully optimized by focusing solely on static posture or rigid structural models. True balance and fluidity arise from dynamic adjustments—the pre-movement shifts of the center of gravity that prepare the body for motion.
By incorporating these principles into hardware design and control algorithms, robots will be able to:
- Initiate movements with greater ease and efficiency.
- Adapt to external disturbances (like sudden obstacles) more naturally.
- Exhibit balance capabilities that mirror, or even surpass, those of humans.
Understanding these subtle preparatory actions will allow us to overcome the current challenges in robotic mobility and take a significant step toward creating robots with human-like balance and responsiveness.
Reflections on Optimizing Robotic Balance Through Hardware Design (Part 4)
When observing the design of modern humanoid robots—including high-end or even toy robots—it’s hard to overlook a recurring issue: many of the main movable joints and actuators (motors) in their torsos and necks are positioned along the central axis.
This design choice seems logical on the surface because it mimics the location of the human spine and other midline bones like the neck, jaw, and nasal bridge. However, from a muscular perspective, this is an oversimplification and potentially inefficient.
The Role of Muscles vs. Bones in Human Movement
- Bones :
- Bones form a central, stable support structure—like the human spine.
- They are positioned along the midline but are not responsible for movement. Instead, they act as a frame or anchor.
- Muscles :
- Unlike bones, muscles are distributed symmetrically on both sides of the body’s midline.
- Their primary role is to pull, bend, or flex the central skeletal structure, enabling efficient movement.
- This off-center distribution allows muscles to apply force more effectively using torque (rotational force).
For example:
- If you imagine a gorilla pulling on a slender tree to bend it, the gorilla pulls from the side—not directly on top of the tree. This is far more efficient than trying to generate force along the centerline.
Problems with Current Robot Design
- Centralized Motors :
- Many humanoid robots place motors or actuators directly along the torso’s midline.
- Whether for vertical twisting (waist rotation) or horizontal bending (forward/backward flexion), this design mimics “tying the gorilla directly to the tree trunk.” The motor must work harder to create movement.
- Rigid Torsos :
- Many humanoid robots feature torsos that are effectively single rigid blocks—like a refrigerator.
- With such a design, the robot’s balance is overly reliant on complex computations to prevent tipping over.
- This “brute-force stability” comes at the cost of computational inefficiency and energy waste.
In contrast, humans rely on dynamic balance adjustments achieved through off-center muscular forces, which are naturally energy-efficient and adaptable.
A Torque-Based Perspective for Robot Design
To address these issues, I propose a redesign based on principles of torque and muscular mimicry:
- Off-Center Motors :
- Place actuators (motors) on both sides of the midline to mimic how human muscles pull and flex the skeletal structure.
- This design reduces the energy needed to generate rotational motion (twisting or bending) and improves balance control.
- Segmented Torsos :
- Replace rigid, single-block torsos with segmented structures that more closely resemble the human spine.
- A segmented design would allow for micro-adjustments in posture and balance, reducing the need for complex real-time computations.
- Dynamic Weight Shifting :
- Incorporate mechanisms to allow dynamic adjustments in weight distribution, similar to how human muscles shift the center of gravity.
- This would enable robots to adapt to uneven surfaces and unexpected disturbances with greater agility.
The Cost of Inefficiency
When I see robots with:
- A solid, block-like torso, or
- Motors placed directly along the midline,
I can’t help but think that these robots are inherently inefficient. To prevent them from falling over, designers must rely on:
- Excessive computational power to maintain balance.
- Overcompensating with complex software algorithms to control movements.
This approach might work, but it feels unnecessarily cumbersome. Why not adopt a more energy-efficient and biomechanically inspired solution?
Conclusion: Toward Smarter, More Human-Like Robots
Humans maintain balance and movement not through brute force or rigid structures, but through:
- Off-center muscle forces.
- Segmented, flexible skeletal structures.
- Dynamic weight shifting to adjust the center of gravity.
By applying these principles to robotic hardware design—particularly in the torso region—we can create robots that:
- Move more fluidly and efficiently.
- Maintain balance with less computational overhead.
- Adapt naturally to their environment, much like humans do.
In the end, the goal is to create robots that don’t just stand and move like humans in appearance, but that replicate the ingenious mechanical efficiency of human movement.