DirectHop: a 1-gram robot that controls jump height and self-rights after landing

DirectHop demonstrates a different approach to miniature hopping robots. Rather than using stored elastic energy in springs and complicated latches, the design uses a very small electric motor to power each jump directly. That allows precise control of jump height and simplifies the mechanical architecture at the millimeter scale.

The robot weighs roughly 1 gram and fits in the palm of a hand. A tiny motor is mounted on a vertical tower and connected by fishing line. When the motor accelerates and spools the line, it lifts itself up the tower and pulls the robot body into the air. Three folding legs keep the motor aligned with the robot's foot during launch.

Instead of an all-or-nothing spring discharge, the motor approach varies jump energy by changing current to the motor. The researchers report single-centimeter accuracy in jump height by adjusting motor input. That continuous control simplifies planning for tasks like clearing a stair step or reaching a specific observation point.

Robots that hop must often be reset after landing. DirectHop includes a roll cage to survive landings on its side. To right itself, the robot reels the motor back down the tower, shifting its center of gravity and rolling the body back onto its foot. The team achieved roughly 90% success at righting this way, enabling multiple hops in sequence without manual intervention.

The prototype relies on wired power and currently lacks deliberate steering or precise orientation control. Its righting motion is a largely open-loop twitch that returns the robot upright most of the time but does not offer controlled rotation. Without onboard power or navigation, the system is still a lab prototype rather than a field-ready device.

The team described several concrete next steps: add onboard solar cells and a battery for untethered operation, include a vibration motor or retractable feet to permit steering and controlled spin, and integrate an onboard camera and processor for navigation. The stated goal is autonomous stair climbing: the robot would identify a step with a camera, calculate required jump height, steer into position, jump, self-right, and repeat.

The researchers suggest that because the individual components are small and inexpensive, a fully featured version could be produced at low cost—one estimate mentioned by the team was about $10 per unit. Low-cost, semidisposable hopping robots could be deployed in large numbers to carry out distributed tasks such as monitoring infrastructure, agricultural sensing, or exploring difficult terrain where losing some units is acceptable.

Key areas to validate in future work include reliable untethered power, repeatable steering and orientation control, integration of sensing and autonomy, and real-world endurance across varied surfaces and obstacles. Demonstrating consistent stair-climbing with onboard perception would be a practical milestone toward field use.