Nobody wins alone in a room full of robots
Field notes from a 无人之境 salon: fire robots, deep-sea verification tables, space embodied AI — and a LinkerBot hand passed around before any pitch deck.
Reel · 30s · Nobody wins alone
Field notes from a 无人之境 ("no man's land") salon hosted by Mars Accelerator. First person, by Christian.
I held a robot hand last night. Five metal fingers, rubber pads, a brass gear at the base of the thumb, and LINKERBOT printed on the heel. It fit in one palm. Someone passed it to me before anyone passed me a pitch deck.
That order — hardware first, claims second — set the tone for the whole evening.
The room
The salon was called 无人之境 — no man's land. The agenda read like a cross-section of physical AI: Tongji's Robotics & Multi-Body Systems Lab, Final Sequence, Mingti, Gramman fire equipment, HRSTEK 合时, Hanjie, Heiman, Chaonao Weilai and SJTU Boyuan. Startups, labs, investors.
Nobody pitched "we win the category alone."
The hardware parade
HRSTEK 合时 opened with product video: a tracked robot working in rubble, a water-cannon shell live-fire demo (水炮弹实战射击演示), and a small explosive-ordnance-disposal robot. (Claimed — company video; I didn't see a live run.)
Gramman showed fire-truck lines next to a full family of firefighting robots: tracked water cannons, an arm, a quadruped, a UAV, scouts. (Claimed — company slides.)
Hanjie 瀚界 / NAMIN Smart Sonar talked underwater acoustic-optical fusion — getting sound and vision to agree when the water won't cooperate.
Tongji's lab put up a deep-domain operations table: pipeline inspection, underwater welding, shipwreck salvage. What I liked most was the honesty on the slide — parts of it are still marked as needing verification. That is a lab telling you where the proof isn't yet. (Confirmed in the room: the slide said so.)
Mingti 明体科技 asked the strangest, best question of the night: why do embodied AI in space? Their answer was about limits — teleoperation and pre-coded behaviour hit walls off-planet, and space robots lack the data, power and heat budget, and reliability margins that ground robots take for granted. (Company framing.)
And an inspection pitch walked through cases on railways and underground pipes, then power, bridges and tunnels. (Claimed — "verified cases" per the slide.)
Why it matters
Feeds sell lone winners: one logo, one round, one demo video. This room ran on the opposite.
The deep-sea table needs sensors. The fire line needs platforms. The space pitch needs ground data it doesn't have yet. The hand needs someone who builds the arm. Mars Accelerator's own deck framed it as co-creation with enterprises and an ecosystem map, and claimed 100+ community events and 7,000+ attendees. (Claimed — host deck; not independently checked.)
I keep a database for a living: Confirmed vs Reported, fleet counts, who actually ships. Spreadsheet work is lonely. Last night the database had a pulse.
What I'm watching
- Proof in the water. When Tongji's "needs verification" becomes a verified weld or salvage run.
- Fire robots in real incidents, not just demo videos.
- The hand. LinkerBot put hardware in people's palms. I want to see it on someone else's arm.
The story isn't who raised last. It's who shows up in the same room — and still has to prove the machine in the water, on the pipe, in the smoke.
Nobody wins alone. More from that night soon: hardware first, claims second.
Photos: Christian / Behind Robotics. Slides shown belong to the presenting companies.
