Videos: Autonomous Warehouse Robots, Drone Delivery

Video Friday is your weekly choice of superior robotics movies, collected by your mates at IEEE Spectrum robotics. We additionally submit 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

Get pleasure from immediately’s movies!

To coach the following era of autonomous robots, scientists at Toyota Research Institute are working with Toyota Manufacturing to deploy them on the manufacturing unit flooring.

[ Toyota Research Institute ]

Thanks, Erin!

Okay, however such as you didn’t present the actually cool bit…?

[ Zipline ]

We’re introducing KinetIQ, an AI framework developed by Humanoid, for end-to-end orchestration of humanoid robotic fleets. KinetIQ coordinates wheeled and bipedal robots inside a single system, managing each fleet-level operations and particular person robotic conduct throughout a number of environments. The framework operates throughout 4 cognitive layers, from job allocation and workflow optimization to job execution based mostly on Imaginative and prescient-Language-Motion fashions and whole-body management taught by reinforcement learning, and is proven right here operating throughout our wheeled industrial robots and bipedal R&D platform.

[ Humanoid ]

What if a robotic will get broken throughout operation? Can it nonetheless carry out its mission with out speedy restore? Impressed by the self-embodied resilience methods of stick insects, we developed a decentralized adaptive resilient neural management system (DARCON). This method permits legged robots to autonomously adapt to limb loss, guaranteeing mission success regardless of mechanical failure. This progressive strategy results in a way forward for actually resilient, self-recovering robotics.

https://www.youtube.com/watch?v=Bp7esFyYV4g

[ VISTEC ]

Thanks, Poramate!

This animation reveals Perseverance’s perspective throughout a drive of 807 toes (246 meters) alongside the rim of Jezero Crater on 10 December 2025, the 1,709th Martian day, or sol, of the mission. Captured over 2 hours and 35 minutes, 53 navigation-camera (Navcam) picture pairs have been mixed with rover knowledge on orientation, wheel velocity, and steering angle, in addition to knowledge from Perseverance’s inertial measurement unit, and positioned right into a 3D digital surroundings. The result’s this reconstruction with digital frames inserted about each 4 inches (0.1 meters) of drive progress.

[ NASA Jet Propulsion Lab ]

[ Unitree ]

Representing and understanding 3D environments in a structured method is essential for autonomous brokers to navigate and motive about their environment. On this work, we suggest an enhanced hierarchical 3D scene graph that integrates open-vocabulary options throughout a number of abstraction ranges and helps object-relational reasoning. Our strategy leverages a imaginative and prescient language mannequin (VLM) to deduce semantic relationships. Notably, we introduce a task-reasoning module that mixes large language models and a VLM to interpret the scene graph’s semantic and relational info, enabling brokers to motive about duties and work together with their surroundings extra intelligently. We validate our technique by deploying it on a quadruped robotic in a number of environments and duties, highlighting its skill to motive about them.

[ Norwegian University of Science & Technology, Autonomous Robots Lab ]

Thanks, Kostas!

We current HoLoArm, a quadrotor with compliant arms impressed by the nodus construction of dragonfly wings. This design offers pure flexibility and resilience whereas preserving flight stability, which is additional strengthened by the mixing of a reinforcement-learning management coverage that enhances each restoration and hovering efficiency.

[ HO Lab via IEEE Robotics and Automation Letters ]

On this work, we current SkyDreamer, to one of the best of our information the primary end-to-end vision-based autonomous-drone racing coverage that maps straight from pixel-level representations to motor instructions.

[ MAVLab ]

This video showcases AI Employee, geared up with five-finger arms, performing dexterous object manipulation throughout various environments. By teleoperation, the robotic demonstrates exact, humanlike hand management in a wide range of manipulation duties.

[ Robotis ]

Autonomous following, 45-degree slope climbing, and dependable payload transport in excessive winter circumstances, constructed to help operations the place environments push the bounds.

[ DEEP Robotics ]

Dwelling architectures, from vegetation to beehives, adapt repeatedly to their environments via self-organization. On this work, we introduce the idea of architectural swarms: programs that combine swarm robotics into modular architectural façades. The Swarm Backyard exemplifies how architectural swarms can remodel the constructed surroundings, enabling “living-like” structure for practical and artistic functions.