Video Friday is your weekly collection of superior robotics movies, collected by your pals at IEEE Spectrum robotics. We additionally put up a weekly calendar of upcoming robotics occasions for the following few months. Please send us your events for inclusion.
ICRA 2026: 1–5 June 2026, VIENNA
Summer School on Multi-Robot Systems: 29 July–4 August 2026, PRAGUE
Take pleasure in in the present day’s movies!
Human athletes reveal versatile and extremely dynamic tennis abilities to efficiently conduct aggressive rallies with a high-speed tennis ball. Nonetheless, reproducing such behaviors on humanoid robots is troublesome, partially as a result of lack of excellent humanoid motion information or human kinematic movement information in tennis eventualities as reference. On this work, we suggest LATENT, a system that Learns Athletic humanoid TEnnis abilities from imperfect human motioN daTa.
[ LATENT ]
A fantastically designed robotic impressed by Strandbeests.
We consider we’re the first robotics firm to reveal a robotic peeling an apple with twin dexterous human-like fingers. This breakthrough closes a key hole in robotics, attaining bimanual, contact-rich manipulation and transferring far past the boundaries of easy grippers.
Immediately’s AI models (VLMs) are glorious at notion however wrestle with motion. Controlling high-degree-of-freedom fingers for duties like that is extremely advanced, and exact finger-level teleoperation is almost unattainable for people. Our first step was a shared-autonomy system: moderately than controlling each finger, the operator triggers pre-learned abilities like a “rotate apple or tennis ball” primitive by way of a keyboard press or pedal. This makes scalable data collection and RL coaching attainable.
How does the AI handle this? We created “MoDE-VLA” (Combination of Dexterous Specialists). It fuses imaginative and prescient, language, power, and contact information by utilizing a workforce of specialist “consultants,” making management in high-dimensional areas secure and efficient. The mix of those two improvements permits for seamless, contact-rich manipulation. The human offers high-level steerage, and the robotic executes the advanced in-hand coordination required.
[ Sharpa ]
Thanks, Alex!
It was nice to see our identify amongst the opposite “AI Native” corporations throughout the NVIDIA GTC keynote. NVIDIA Isaac Lab helps us practice reinforcement learning insurance policies that allow the UMV to drive, bounce, flip, and hop like a professional.
This Finger-Tip Changer expertise was collectively researched and developed via a collaboration between Tesollo and RoCogMan LaB at Hanyang College ERICA. The challenge integrates Tesollo’s sensible robotic hand improvement expertise with the lab’s experience in robotic manipulation and gripper design.
I don’t know why extra robots don’t do that. Additionally, these pointy fingertips are terrifying.
[ RoCogMan LaB ]
Right here’s an upcoming ICRA paper from the Fluent Robotics Lab on the College of Michigan that includes an operational PR2! With purposeful batteries!!!
This video showcases the sector assessments and interplay capabilities of KAIST Humanoid v0.7, developed on the DRCD Lab that includes in-house actuators. The management coverage was educated via deep reinforcement learning leveraging human demonstrations.
[ KAIST DRCD Lab ]
This wants to come back in grownup dimension.
[ DEEP Robotics ]
I didn’t know this, however apparently shoeboxes are actually annoying to control as a result of if you happen to seize them by the lid, they simply open, so specialised {hardware} is required.
[ Nomagic ]
Thanks, Gilmarie!
This paper presents a technique to recuperate quadrotor Unmanned Air Automobiles (UAVs) from a throw, when no management parameters are identified earlier than the throw.
[ MAVLab ]
Uh oh, robots can see glass doorways now. We’re in hassle.
[ LimX Dynamics ]
This drone hugs timber <3
[ Stanford BDML ]
Electronic waste is likely one of the fastest-growing environmental issues on the planet. As robotics and digital methods turn out to be extra widespread, their environmental footprint continues to extend. On this analysis, scientists developed a completely biodegradable smooth robotic system that integrates digital units, sensors, and actuators, but fully decomposes after use.
[ Nature ]
We developed a distributed algorithm that permits a number of aerial robots to flock collectively safely in advanced environments, with out specific communication or prior information of the environment, utilizing solely on-board sensors and computation. Our method ensures collision avoidance, maintains proximity between robots, and handles uncertainties (monitoring errors and sensor noise). Examined in simulations and real-world experiments with as much as 4 drones in a dense forest, it proved strong and dependable.
[ RBL ]
The College of Pennsylvania’s 2025 President’s Sustainability Prize winner Piotr Lazarek has developed a system that makes use of satellite tv for pc information to pinpoint inefficiencies in farmers’ fields, conducts real-time soil evaluation with autonomous drones to grasp why they happen, and generates exact fertilizer utility maps. His startup Nirby goals to extend productiveness in farm areas which can be underperforming and scale back fertilizer in high-performing ones.
[ University of Pennsylvania ]
The manufacturing model of Atlas is a departure from the everyday humanoid kind issue, favoring industrial utility over human likeness. Supposed for purposeful work in an industrial setting, Atlas has a kind issue that alerts its function as a machine moderately than a companion or pleasant assistant. Be a part of two lead {hardware} engineers and our head of business design for a technical dialogue of how key product necessities, starting from passive thermal management to a modular structure, dictated a daring new imaginative and prescient for a humanoid.
[ Boston Dynamics ]
Dr. Christian Hubicki offers a chat exploring the widespread themes of contemporary robotics research and his time on the truth competitors present, Survivor.
From Your Website Articles
Associated Articles Across the Internet
