What we’re watching next in humanoids

Live humanoid demos on Tokyo's expo floor promise the year’s sharpest test of hype versus hardware.

TechCrunch is packing its bags for Tokyo, headed to SusHi Tech 2026, where AI, Robotics, Resilience, and Entertainment collide. The event is pitched as a proving ground for the next wave of humanoid robots, with live demonstrations intended to cut through the marketing noise that routinely surrounds “revolutionary” prototypes. In short: show us the gait, the grip, and the battery life—then show us the data that backs it up.

The report framing the trip notes that SusHi Tech 2026 will feature humanoid demonstrations, but it does not publish model names, DOF counts, or payload specifications for any specific robot. That absence is telling: in a field where press footage can be compelling, the numbers still matter. In humanoid robotics, the devil is in the details—how many degrees of freedom move the ankle, knee, hip, or wrist; how much weight a hand can reliably lift without slipping; how much torque a single actuator can deliver before control becomes brittle. The TechCrunch piece functions as event-level coverage, not a capex sheet for a single platform. As a result, this article can only discuss trends and implications rather than report on discrete, verifiable specs for a particular model.

Engineering documentation shows a growing consensus: teams are converging on more capable but still brittle stacks—better actuation, tighter control loops, and more integrated perception, all bound by the realities of power, heat, and safety. Lab testing confirms steady gains in manipulation and locomotion, but field readiness remains a separate milestone. The challenge isn’t just making a humanoid stand up; it’s making it stand up, walk, reach, and hand you an object—consistently and safely—under real-world lighting, acoustics, and human interaction. The lack of disclosed DOF and payload data suggests that the event will emphasize demonstrations of capability over publication of performance envelopes.

Looking ahead, the SusHi focus areas imply an emphasis on end-to-end capability: perception translating into precise, safe manipulation; locomotion across varied surfaces; and robust interaction with people in shared spaces. Expect attendees to scrutinize how teams balance power budgets with runtime, how control software handles cluttered environments, and how quickly a robot can recover from a slip or mis-grasp. In past cycles, the most telling improvements have come not from one flashy feature but from the reliability of the entire loop: sensing, planning, control, and actuation working in near-unison.

What we’re watching next in humanoids

DoF and payload disclosures: will demonstrations publish concrete numbers or remain high-level?

Field readiness signals: any plan shown for operation outside controlled demo floors, including safety protocols?

Power, runtime, and charging: are there multi-hour operational modes or only quick demos with tethered power?

Perception-to-action latency: how fast do the robots sense, decide, and act in dynamic human environments?

Manipulation reliability: prosthetic-like grasping durability across varied objects and textures.

Source

TechCrunch: TechCrunch is heading to Tokyo — and bringing the Startup Battlefield with it. https://techcrunch.com/2026/04/10/techcrunch-is-heading-to-tokyo-and-bringing-the-startup-battlefield-with-it/

Sources

TechCrunch is heading to Tokyo — and bringing the Startup Battlefield with it