Video Friday: Drone Squeezes Itself to Fit Through Tight Spaces (Nov 20). IEEE Spectrum shows a quadrotor prototype that narrows from 31.3 cm to 14.1 cm as its side arms bend on contact, allowing it to pass through a 15 cm gap. Passive compliance could help small inspection drones pass through openings narrower than their normal flight profile.1
Readiness (Research prototype). The cited team demonstrates a physical prototype flying through a constrained opening.
Pros. The side arms fold on contact without a separate folding action.
Cons. The demonstration does not establish durability over repeated impacts or reliability across irregular openings.
Watch the quadrotor’s two side arms fold on contact as it flies through the 15 cm opening.
Ferrofluid Shape Control with Deep Reinforcement Learning (Nov 20). A ferrofluid-control preprint uses camera feedback and one fixed electromagnet to guide a magnetic fluid into transient shapes. The setup tests whether learned control can produce useful deformation with very little actuation hardware.2
Readiness (Research prototype). The authors trained through physical interaction and repeatedly produced shapes including asterisks and polygons.
Pros. One electromagnet sufficed for several shapes in the lab setup.
Cons. The study does not establish load handling or repeatability outside that setup.
Stereo Cameras
VB1943 and robotics: When AI is in the eye (or CMOS sensor) of the beholder (Nov 18). ST describes a Leopard Imaging stereo setup built around 5-megapixel sensors and a Jetson AGX Orin developer kit, and discusses its industrial VB1943 counterpart. For robots that rely on dense depth, the write-up highlights the need to evaluate both optics and online calibration.3
Readiness (Research prototype). The study cited by ST used a module with the automotive VB1940 variant, while ST discusses the VB1943 for industrial designs.
Pros. The reference setup gives teams a starting point for testing depth estimation and calibration drift.
Cons. The studied setup does not establish equivalent results with the VB1943 under a target robot’s vibration and temperature conditions.