Robot Videos: Atlas Humanoid, CES 2026 Bots , and More

Video Friday is your weekly collection of superior robotics movies, collected by your mates at IEEE Spectrum robotics. We additionally submit a weekly calendar of upcoming robotics occasions for the subsequent few months. Please send us your events for inclusion.

ICRA 2026: 1–5 June 2026, VIENNA

Get pleasure from immediately’s movies!

[ Boston Dynamics ]

[ Roborock ]

Wow, SwitchBot’s CES 2026 video present virtually as many robots of their fantasy house as I’ve in my actual house.

[ SwitchBot ]

What is going on in robotics proper now that I can derive extra satisfaction from watching robotic process automation than I can from watching yet one more humanoid video?

[ ABB ]

Sure, that is positively a robotic I would like in shut proximity to my life.

[ Unitree ]

The video beneath demonstrates a MenteeBot studying, by means of mentoring, how one can substitute a battery in one other MenteeBot. No teleoperation is used.

[ Mentee Robotics ]

Personally, I feel that we must always encourage humanoid robots to fall far more typically, simply in order that we are able to see whether or not they can rise up once more.

[ Agility Robotics ]

Attaining long-horizon, dependable clothes manipulation in the true world stays some of the difficult issues in robotics. This dwell check demonstrates a powerful step ahead in embodied intelligence, vision-language-action programs, and real-world robotic autonomy.

[ HKU MMLab ]

Tens of millions of individuals world wide want help with feeding. Robotic feeding programs provide the potential to reinforce autonomy and high quality of life for people with impairments and cut back caregiver workload. Nonetheless, their widespread adoption has been restricted by technical challenges comparable to estimating chunk timing, the suitable second for the robotic to switch meals to a person’s mouth. On this work, we introduce WAFFLE: Wearable Method For Feeding with LEarned Chunk Timing, a system that precisely predicts chunk timing by leveraging wearable sensor information to be extremely reactive to pure person cues comparable to head actions, chewing, and speaking.

[ CMU RCHI ]

Humanoid robots at the moment are out there as platforms, which is a good way of sidestepping the entire practicality query.

[ PNDbotics ]

We’re introducing Spatially-Enhanced Recurrent Items (SRUs) — a easy but highly effective modification that permits robots to construct implicit spatial reminiscences for navigation. Printed within the Worldwide Journal of Robotics Research (IJRR), this work demonstrates as much as +105% enchancment over baseline approaches, with robots efficiently navigating 70+ meters in the true world utilizing solely a single forward-facing digicam.