EN
Search
Photo: MIT News | Massachusetts Institute of Technology

MIT researchers build flapping-wing microrobot that flies with its own onboard sensors and computers

MIT researchers have built a flapping-wing microrobot that carries its own sensors and computing electronics. According to a paper published in Science Robotics, the robot can follow a complex path and land on a sunflower outside the lab's motion-capture setup.

A tiny flying robot rests on a sunflower. Photo: MIT News | Massachusetts Institute of Technology

A tiny flying robot rests on a sunflower. Photo: MIT News | Massachusetts Institute of Technology

Over the past decade, the group has developed small aerial microrobots that are fast, agile and durable, with possible uses in precision agriculture and search-and-rescue. Earlier versions could not lift the electronics needed for navigation, so they stayed connected to a motion-capture system in the lab.

“This is the first demonstration of an aerial microrobot that is sensory and compute autonomous.”

Kevin Chen, associate professor in the Department of Electrical Engineering and Computer Science (EECS)

The robot is still linked to an offboard power source, but the researchers said the work points to future use in farming. They described flying microrobots that could move through almond or blueberry plantings to pollinate flowers in areas where wild pollinator populations have declined because of habitat loss. Such robots could also hover near crops to check plant health and spot pests early.

To make room for electronics, the team changed the artificial muscles that drive the wings. The team's earlier soft-actuator microrobot had four wings, each moved by soft actuators. In this work, the researchers enlarged the actuators by 50 percent and adjusted their radius and length, which raised net lift by 46 percent.

The robot has four wings and is shaped like a plus sign. Photo: MIT News | Massachusetts Institute of Technology

The robot has four wings and is shaped like a plus sign. Photo: MIT News | Massachusetts Institute of Technology

The researchers reported a lift-to-weight ratio of 1.9, enough lift to carry a custom electronics package that weighs 244 milligrams. The package holds the sensors and a microcontroller needed for onboard sensing and computation. The sensor set includes an inertial measurement unit to estimate attitude, an optical flow sensor for lateral speed and position, and a time-of-flight sensor for height.

The robot is on the left and six tiny pieces of the computer are on the right. Photo: MIT News | Massachusetts Institute of Technology

The robot is on the left and six tiny pieces of the computer are on the right. Photo: MIT News | Massachusetts Institute of Technology

Without the precision of a motion-capture system, the robot cannot determine its absolute position and drifts off course. To address this, the team wrote a sensing algorithm that estimates orientation, angular velocity and position very quickly, using simplified calculations to reduce computing load. The researchers found that updating position continuously at a faster rate performed better than a more accurate but slower estimate.

In flight tests, the robot hovered at different heights and followed point-to-point paths, with position errors of only a few centimeters. A joystick sent high-level position commands that guided it around several plants before it landed on a sunflower. The researchers said that task is difficult even with a motion-capture system.

The team measured the vehicle's power use and found the electronics consume only about 6 percent of the total, with the rest powering the artificial muscles and wings. The researchers also identified ways to cut the electronics' power use further, which could leave room for added components such as a camera.

The researchers are now pursuing two goals: adding a camera for autonomous object recognition and obstacle avoidance, and integrating power electronics and batteries so the robot could fly untethered outdoors. The work was funded by the U.S. National Science Foundation, the U.S. Office of Naval Research, the U.S. Air Force Office of Scientific Research, MathWorks and a Zakhartchenko Fellowship.

Source