A Challenge to Roboticists: My Humanoid Olympics
Why Humanoid Robot Competitions Must Move Beyond Entertainment
China’s recent World Humanoid Robot Games made headlines for its spectacle—think real-life Rock ‘Em Sock ‘Em Robots. But beneath the surface excitement, a growing chorus within robotics wonders: When will we see robots compete at the tasks that matter most to people? The answer isn’t in flashy brawls but in robots tackling the everyday challenges that fill our lives—folding laundry, opening doors, handling tools, and cleaning up messes.
The viral popularity of robot laundry-folding videos is no accident. These clips fascinate not because they’re high-octane, but because they make us imagine a world where robots handle the mundane chores we’d rather avoid. However, as impressive as these feats seem, they reveal not mastery, but the limits of current AI and robotics. To truly bring robots into our homes and workplaces, the field must push past today’s sweet spots and into genuinely useful, general-purpose manipulation.
Below, we outline a set of “Humanoid Olympics” events—challenges designed to stretch the state of the art in robotic manipulation, inspired by Benjie Holson’s call for progress. Each event reflects a common human task that, if achieved by robots, would mark a leap forward for real-world utility.
Where Robots Stand Today: The State of Robotic Manipulation
Nearly all major leaps in robot manipulation over the past few years rely on “learning from demonstration.” This typically means a human operator—sometimes using a duplicate robot or a virtual reality interface—physically guides the robot through a task, often hundreds of times, to generate training data. A neural network then learns to mimic these actions.
This approach has powered breakthroughs in seemingly chaotic or complex tasks, like pulling the corner of a towel flat or unbunching laundry. However, it also exposes serious limitations:
- No force feedback at the wrists: Human operators can’t “feel” what the robot feels, making it hard to fine-tune grip or detect slippage.
- Limited finger control: Even the most dexterous robots rarely match the subtlety of human hand movements; most manipulation is still “open or close” rather than nuanced.
- No sense of touch: Human hands are laden with tactile sensors; robots, even those with advanced in-finger vision, lag far behind.
- Medium precision: Robots still struggle with tasks requiring sub-centimeter accuracy, especially if the workspace is cluttered or the objects are deformable.
Despite these challenges, robots have begun to master some tasks that were out of reach even a few years ago. But the gap between “can demonstrate” and “can generalize and reliably perform” remains wide.
The Humanoid Olympics: Five Events That Matter
Event 1: Doors
Why It Matters
Doorways are ubiquitous in human environments. If a robot can’t open and navigate through a variety of doors, its usefulness in real-world settings is limited.
- Bronze: Entering a round-knob push door. This task is near current capabilities, but requires precise force and trajectory.
- Silver: Entering a lever-handle, self-closing push door. The added force and timing challenge most control systems.
- Gold: Entering a lever-handle, self-closing pull door. The “boss fight”—requiring either rapid dynamic movement or coordinated use of both arms to keep the door from re-closing.
Technical Challenge: These tasks require robust whole-body coordination, real-time grip adjustment, and the ability to handle asymmetric and dynamic forces—all without dropping or losing control of the handle.
Event 2: Laundry
Why It Matters
Laundry folding is a perfect storm of deformable object manipulation, multi-step planning, and two-handed coordination. It represents the kind of real household chore most people want help with.
- Bronze: Fold an inside-out T-shirt. Achievable with current learning-from-demonstration approaches, but still nontrivial.
- Silver: Turn a sock inside-out. Requires inserting a hand and pinching fabric—demanding precision grip and dexterity.
- Gold: Hang a men’s dress shirt on a hanger, starting with one sleeve inside-out and unbuttoned. Must finish with shirt properly hung, sleeve fixed, and at least one button fastened—a task 3–10 years away, given the need for small, strong, highly dexterous hands.
Technical Challenge: Deformable object manipulation is still a frontier in robotics. Even folding reliably, let alone unbunching sleeves or buttoning, demands new perception and control algorithms.
Event 3: Tools
Why It Matters
Much of human work is done with tools. A robot that can use tools reliably can help with cleaning, cooking, repairs, and more.
- Bronze: Spray window cleaner and wipe with paper towels. Requires grasping, actuating a trigger, and cleaning motions.
- Silver: Make a peanut butter sandwich, including scooping and spreading. Picking up, adjusting grip, and applying force with a knife is a major leap.
- Gold: Select a key from a keyring, insert, and turn it in a lock—without putting the keys down. Demands in-hand manipulation and very fine motor control.
Technical Challenge: Tool use combines grip strength, dexterity, and in-hand object reconfiguration—areas where even leading robots like Tesla’s Optimus or TRI’s research bots face hurdles.
Event 4: Fingertip Manipulation
Why It Matters
Humans routinely manipulate small objects in the hand—think rolling socks, separating a bag, peeling fruit. These actions combine force, precision, and tactile feedback.
- Bronze: Roll matched socks together. Dexterity and precision, but low force.
- Silver: Separate and tear off a dog poop bag from a roll. Requires sliding motion between fingertips and controlled tearing.
- Gold: Peel an orange, by hand. High force, high precision, and continuous adjustment.
Technical Challenge: Robots lack the dense tactile sensing and motor control that makes these actions trivial for humans. Progress here would open doors to countless delicate tasks.
Event 5: Wet Manipulation
Why It Matters
Real-world chores are often messy or involve liquids. Robots must overcome the aversion to water and slippery, deformable objects if they are to clean kitchens or bathrooms.
- Bronze: Wipe a countertop with a sponge. Involves water risk and precise control to avoid soaking the robot.
- Silver: Clean peanut butter off the robot’s own manipulator. A meta-cleaning task, essential for long-term autonomy.
- Gold: Wash a greasy pan in a sink. Combines water, soap, grease, and the need for thorough scrubbing.
Technical Challenge: Wet environments introduce electrical, mechanical, and control complexities. Grasping wet, slippery, or greasy items without losing them is a major unsolved problem.
Why This Matters: Real-World Impact and Implications
Progress on these “Olympic” events isn’t about entertaining competitions—it’s about unlocking the true value of robotics for society. Imagine a robot that can reliably open doors, fold laundry, make a sandwich, fetch your keys, and clean your kitchen. Such a machine would revolutionize elder care, support people with disabilities, and free up human time for higher-value pursuits.
Yet, each event exposes a technical chasm: we need advances in tactile sensing, force feedback, dexterous manipulation, robust perception, and generalizable learning. Today’s learning-by-demonstration pipeline is powerful but brittle; it struggles to generalize to new objects, settings, or slight variations in tasks.
The practical insight for roboticists and AI engineers is clear: focus on the “last mile” tasks that make robots genuinely useful in homes and businesses. For investors and entrepreneurs, the message is that the biggest economic impact will come not from robots that can walk or fight, but from those that can do dishes, pick up toys, or sort laundry.
Looking Ahead: The Road to Truly General-Purpose Robots
The Humanoid Olympics challenge is not just a wishlist—it’s a roadmap for progress in general-purpose robotics. As researchers begin to solve these milestones, we’ll see a shift from specialized, single-task machines to flexible, helpful assistants. This will demand new approaches in machine learning, hardware design, and sensor integration.
Ultimately, the field’s success will be measured not by who wins the next robot boxing match, but by how soon we see robots that can tackle the chores and challenges of everyday life, side by side with humans. As these Olympic events are conquered, the dream of practical, affordable, and widely useful robots will finally move from science fiction to reality.
