- Published on
Stepping Forward: The Debate Over Active vs. Passive Toes in Humanoid Robotics
- Authors
- Name
- The Editorial Team
- @humanoidsdaily
The Footwork Debate: Articulated Toes on Optimus Spark Discussion on Humanoid Agility
Recent videos of Tesla's Optimus robot showcasing impressive dance-like maneuvers have brought a seemingly small detail into the spotlight: its feet. While the agility displayed is a testament to advancements in simulation and reinforcement learning, the design of Optimus's articulated toes has become a point of discussion among robotics experts and enthusiasts. This raises broader questions about the optimal approach to foot and toe design in humanoid robots, particularly the trade-offs between passive and active systems.
Optimus: Articulated, Not Necessarily Fully Actuated Toes
Observations of Optimus, particularly in newer videos, have led to speculation about the nature of its toe joints. Robotics expert Scott Walter, in a thread on X and in an episode with Dr. Know-it-all Knows it all on YouTube, clarified that Optimus appears to have an articulated toe section with a hinge joint and a spring mechanism, rather than a fully actuated (motor-driven) one. This means the toe section can bend and flex, returning to a neutral position via the spring, but isn't independently powered to the same degree as other joints.
Walter explained that the noticeable "flop" or upward movement of the toe box seen when Optimus drags its foot backward is likely a result of physics rather than active actuation. As the ankle flexes and the foot slides, the hinge design and ground forces can cause the toe section to lift and appear to move independently. Tesla had mentioned an "articulated toe box" as early as the Gen 2 reveal in December 2023.
A Big Flap Over a Flap Articulated. NOT Actuated. Full Stop. Why? Physics 🧵
The primary benefit of an articulated toe box, even if passive, is to allow for more natural foot-to-ground interaction. A flat, rigid foot, when tilted even slightly, can lose surface contact and pivot on an edge, leading to instability and slippage, especially during the push-off phase of walking. An articulated toe allows for better surface contact as the ankle angle changes, improving grip and stability.
Active vs. Passive Toes: A Balancing Act
The discussion around Optimus's feet highlights a key design choice in humanoid robotics: whether to incorporate active or passive toe joints.
Active toe joints are equipped with their own actuators (motors), allowing for independent and precise control of toe movement. This can enhance propulsion, improve balance, and allow for better adaptation to uneven terrain. Academic studies have shown that active toe joints can contribute to a stronger thrust vector during the "toe-off" phase of walking, potentially enabling faster and more energy-efficient locomotion, and a more human-like gait. They can also increase the effective area of the support polygon during certain movements, enhancing controllability. However, active toes add complexity, weight, and increased energy demands to the robot's design.
Passive toe joints, like the one seemingly employed by Optimus, lack dedicated actuators and rely on built-in flexibility, springs, or the robot's overall dynamics to respond to external forces. These systems are simpler, lighter, and more energy-efficient from a component perspective. However, they offer limited functionality compared to active systems and their effectiveness is heavily dependent on the overall design of the robot's leg and ankle. Some research has explored modeling passive toe joints as torsional springs, allowing natural dynamics and optimization frameworks to govern their motion.
The choice between active and passive systems often depends on the robot's intended applications, with active joints being more suitable for complex tasks and varied environments, while passive joints are favored for simplicity and efficiency in less demanding scenarios.
Industry Approaches: A Mixed Bag
Tesla's approach with Optimus's articulated toes, if indeed primarily passive with spring return, contrasts with some but aligns with other trends in the industry:
Boston Dynamics' Atlas, known for its dynamic capabilities, has historically used a rigid, curved sole, relying on powerful ankle actuation for many of its complex maneuvers.
Figure AI's Figure 02, Apptronik Apollo and Unitree H1 largely depend on ankle actuation combined with soft rubber pads on their feet. These designs prioritize simplicity and robustness in the foot itself, placing more emphasis on the ankle joint's capabilities.
Above: Video showing Apptronik Apollo's rigid feet.
While simpler foot designs reduce complexity at the extremity, they may place greater demands on ankle and other leg actuators and could be less energy-efficient during long strides or when mimicking human-like push-off. The human foot, with its complex structure of bones, ligaments, and muscles, provides significant energy return and adaptability during locomotion, something robotics engineers are still striving to replicate efficiently.
The Path Forward: Efficiency and Agility
The "flap over a flap" surrounding Optimus's toes underscores the nuanced engineering challenges in creating truly human-like bipedal locomotion. While the rapid advancements in control algorithms driven by reinforcement learning are enabling impressive feats of agility, the underlying mechanical design of joints, including those in the feet, remains crucial.
Achieving a balance between complexity, capability, and energy efficiency is paramount. Whether future humanoids will predominantly feature sophisticated, actively actuated toes or rely on clever passive designs augmented by powerful ankle and leg control is an ongoing area of development. As robots are tasked with navigating increasingly complex human environments, the design of their feet will undoubtedly continue to be a critical area of innovation.
See the episode of Dr. Know-it-all Knows it all below:
More reading: