AeRove is a compact drone that turns the guards surrounding its propellers in flight into wheels after landing, allowing it to roll along pipelines. It normally travels on the pipe and takes off only when it needs to hop over an obstacle.

A spring-loaded mechanism switches between flight and ground configurations in about 200 milliseconds. The researchers report roughly 14 times lower current draw while rolling than during continuous flight, and demonstrated straight and curved traversal, obstacle hopping, and simulated leak-marker detection on a 51 cm steel pipe. It was developed by NJIT researchers Caleb Polillio and Petras Swissler.

NYU Tandon School of Engineering's WorMa is a worm-like robot that adapts to land and water by pumping internal water between tanks in its head and tail, shifting its center of mass. The 50 cm, roughly 1 kg robot has five links, with a central pump moving about 300 g of water.

Moving the water forward improved traction on slopes, while shifting it toward the tail made swimming faster. To climb steps, the robot moved the water from head to tail and back again while changing posture, allowing it to clear steps up to 15 cm high.

The Black Thunder Driven Development device monitors token usage in AI development tools such as Claude Code and automatically dispenses Black Thunder chocolate bars when usage reaches set thresholds. An indicator shows progress against the five-hour usage limit, with a full limit corresponding to five bars.

When a reward is triggered, the machine cuts open the wrapper and presents the chocolate ready to eat. A robot arm also replenishes the stock. The project won the Blackathon hackathon and was built by YS_beaver, Yota Saito, Mikihiro Urano, and Tomoki Banno.

RoboSwap is a prototype in which a single robot arm can switch between tools such as grippers and pipettes for different laboratory tasks. The first working version picks up an electronic pipette, aspirates liquid, and transfers it between containers.

The maker does not have a laboratory, so the prototype is tested in a kitchen using soy sauce and saucers. The pipette also lacks automated tip ejection, so the robot presses its ejector against a stool to remove the tip. The system was prototyped by RoboDrop.

Researchers at the University of Waikato deliberately built a robot called Spike using traits that previous research associated with creepiness: an appearance that is hard to classify as human, animal or machine, plus spikes, unusual facial features, dark colors and unpredictable motion.

They showed Spike to 104 participants. Some described it as angry or unsettling, while others thought it was cool or even cute. The team concluded that there may be no universally creepy robot because perceptions vary strongly between people.

Researchers at ETH Zurich's Soft Robotics Lab taught a robotic hand to use its five fingers as legs and locomote on its own. It moved untethered across 14 surfaces, recovered from falls, and supported its own weight while pressing keyboard arrow keys, scoring 29 correct presses out of 32.

The experiments used an off-the-shelf WUJI hand fitted with a Raspberry Pi Zero 2 W, IMU and battery, with control policies trained in Isaac Lab and transferred to the physical system.

Matt Denton's Spoolbot looks almost exactly like a 2 kg spool of 3D-printer filament, complete with real filament wrapped around the outside. At a glance, there is little to suggest that it is a mobile robot.

Inside are an ESP32-C3-based controller, IMU, RC receiver, geared DC motors, and battery packs used as ballast. The spool's black end discs act as the wheels, while the internal frame stays upright as the shell rolls around it.

Maker heychaostheory built a small wheeled mouse droid with a 3D-printed body, ultrasonic obstacle avoidance, an Arduino Uno, motor control, and flashing LEDs. It can drive around on its own while avoiding objects.

Then there is the extra feature: a dedicated party-mode button. Pressing it triggers music, turning a practical autonomous robot into a mobile party machine on command.

After building a fully functional 7:1-scale Arduino UNO, UncleStem made a matching 7:1-scale version of the classic beginner turtle robot. The oversized parts mimic an UNO, L298N driver, ultrasonic sensor and jumper wires, while the internals are replaced with appropriately beefy hardware such as 12–24 V ride-on-toy motors and a pair of 300 W motor drivers.

The giant robot still behaves like the original classroom build: it pans an ultrasonic sensor, detects obstacles and drives around them.

A demonstration about the impact of AI and robotics on jobs put roughly 30 robots on the pavement alongside its message. On September 7, humanoids and quadruped robots appeared outside Poland’s Ministry of Digital Affairs in Warsaw, carrying flags and moving in circles.

The machines were not independently protesting. The event was staged by Democratism, an initiative seeking public debate over automation and the labor market. Chants were played through loudspeakers, and robots including an AgiBot A3 operated by Delta Robots took part.

Poland’s digital affairs minister also met the organizers. The visual argument was unusually literal: to warn that robots may replace human work, the organizers recruited robots to perform the protest itself.

At Adventure World in Wakayama, a robotic penguin swims beside real penguins, moving its flippers to turn and dive.

The robot was built by the park operator’s Kokorobo team. The result is a zoo exhibit where a machine can briefly blend in with the animals people came to see.

The runway at IFA 2026 featured humanoids and quadruped robots rather than human models. During the September 5 “Robots on the Runway” event, machines walked the stage, struck poses, danced and performed moves including backflips.

IFA’s official program listed participants such as Primebot Robots, EngineAI, Dobot, DEEP Robotics, AgiBot and Unitree. Instead of limiting the machines to booths and specification sheets, the event presented their sensing and motion-control abilities as a live runway show.

A space normally built around clothes and human presentation thus became a robotics demonstration. Once machines acquired humanlike bodies, apparently one of the next tasks waiting for them was learning how to make an entrance.