XPENG Shows IRON Switching Between Handshakes, Fist Bumps and High-Fives

XPENG has shared a new IRON demonstration showing its humanoid switching between handshakes, fist bumps and high-fives as a person changes gestures mid-movement. The company says its VLA 2.0 technology lets the robot interpret visual cues and adjust its movements without spoken commands.

Key Takeaways

Show all

• IRON follows changing hand gestures, including switching arms to reach a high-five.
• XPENG says the robot re-evaluates what it sees every 0.2 seconds using VLA 2.0.
• The footage is labeled an R&D test-version demonstration, with no success-rate or end-to-end latency figures disclosed.

In the video posted by XPENG, a tester offers a handshake, changes to a fist bump, then moves into a high-five. When the target moves beyond the reach of the arm IRON initially uses, the robot responds with its other hand.

The accompanying explanation describes a continuous cycle of observation, motion planning and execution. Rather than waiting for one gesture to finish before responding to the next, XPENG says IRON keeps updating its movement as the person's hand changes position. Its stated interval of 0.2 seconds corresponds to five reasoning updates per second; it is not a measurement of the complete time between seeing a gesture and responding.

XPENG's post describes IRON as predicting a person's next move. The concrete behavior presented here is narrower: adapting to changing visual cues during an interaction. The video does not provide a separate test of prediction accuracy or disclose how many attempts were filmed. It also does not establish whether the VLA 2.0 system is identical to the model used in XPENG's vehicles.

CEO He Xiaopeng framed the demonstration as a step toward everyday use: “Natural interaction begins with these simple moments.” He said continued improvements would allow IRON to enter daily life and create value.

The update focuses on a small but useful interaction skill: keeping up when a person changes their mind halfway through a gesture. The on-screen R&D label places it firmly in ongoing development, rather than announcing a production-ready capability or customer rollout.